We are IntechOpen, the world's leading publisher of Open Access books Built by scientists, for scientists

4,000+

Open access books available

116,000+

International authors and editors

120M+

Downloads

Our authors are among the

Top 1%

most cited scientists

12.2%

Contributors from top 500 universities

Selection of our books indexed in the Book Citation Index in Web of Science™ Core Collection (BKCI)

### Interested in publishing with us? Contact book.department@intechopen.com

Numbers displayed above are based on latest data collected. For more information visit www.intechopen.com

## **Meet the editors**

Professor Roberto C. Hincapie received his PhD degree in Telecommunications Engineering from the Universidad Pontificia Bolivariana, Medellin, Colombia in 2009. Currently, he is an Associate Professor at the same university, where he is also the director of GIDATI research group. His current research interests are Wireless Mesh Networks, resource allocation and OFDM wireless

technologies such as WiMAX and LTE. He has taught courses in Traffic Engineering, Simulation and Design of Telecommunication systems. He has participated in several publications in indexed journals and addressed a number of key international conferences.

Professor Javier E. Sierra received his PhD degree in Telecommunications Engineering from the Universidad Pontificia Bolivariana, Medellin, Colombia in 2009. Currently, he is a professor at the Engineering Department of the same university. He works for GIDATI research group and Universidad Pontificia Bolivariana. He has participated in various national and international confer-

ences (IEEE, ACM), and has received awards, including the award for the best paper presented. He is also an author of several publications in indexed journals. His biography has been included in the book Who's Who in the World, 2010 edition, and Who's Who in Science and Engineering by Marquis Who's Who in the World. His research interests include wireless networks, network optimization, traffic grooming and optical transport networks.

Contents

**Preface IX** 

**Part 1 Advanced Transmission** 

Chapter 1 **Hexa-Band Multi-Standard Planar** 

Chapter 3 **A Reconfigurable Radial Line** 

Mohd Faizal Jamlos

Chapter 4 **Reduction of Nonlinear Distortion** 

Chapter 5 **MicroTCA Compliant WiMAX BS** 

Chapter 6 **Space-Time Adaptation and** 

Chapter 7 **Hybrid ARQ Utilizing Lower Rate** 

**Split Architecture with MIMO** 

Cristian Anghel and Remus Cacoveanu

**MIMO Standardization Status 103**  Ismael Gutiérrez and Faouzi Bader

**Part 2 Physical Layer Models and Performance 129** 

Cheng-Ming Chen and Pang-An Ting

Yu-Jen Chi and Chien-Wen Chiu

Chapter 2 **CPW-Fed Antennas for WiFi and WiMAX 19** 

**Techniques, Antennas and Space-Time Coding 1** 

**Antenna Design for Wireless Mobile Terminal 3** 

Sarawuth Chaimool and Prayoot Akkaraekthalin

**in Multi-Antenna WiMAX Systems 59** 

**Slot Array Antenna for WiMAX Application 49** 

Peter Drotár, Juraj Gazda, Dušan Kocur and Pavol Galajda

**Retransmission over MIMO Wireless Systems 131** 

**Capabilities Support Based on OBSAI RP3-01 Interfaces 77** 

### Contents

#### **Preface XI**


X Contents


### Preface

This book has been prepared to present the state of the art on WiMAX Technology. The focus of the book is the physical layer, and it collects the contributions of many important researchers around the world. So many different works on WiMAX show the great worldwide importance of WiMAX as a wireless broadband access technology.

This book is intended for readers interested in the transmission process under WiMAX. All chapters include both theoretical and technical information, which provides an in-depth review of the most recent advances in the field, for engineers and researchers, and other readers interested in WiMAX.

In the first section, *Advanced Transmission Techniques*, readers will find chapters on modern antennas design for future WiMAX communications and the transmission enhancements achieved by space-time coding. In the second section, *Physical Layer Models*, there are several chapters on the Automatic Repeat Request process and the common Peak to Average Power Ratio problem for OFDM modulation. Finally, in the third section the reader will find chapters related to mobile WiMAX problems, handover processes and interaction with other technologies.

> **Prof. Roberto C. Hincapie & Prof. Javier E. Sierra**  Universidad Pontificia Bolivariana, Medellín, Colombia

**Part 1** 

**Advanced Transmission Techniques,** 

**Antennas and Space-Time Coding** 

### **Part 1**

### **Advanced Transmission Techniques, Antennas and Space-Time Coding**

**1** 

*Taiwan* 

**Hexa-Band Multi-Standard Planar Antenna** 

Electronic devices such as mobile phones and laptop computers are parts of modern life. Users of portable wireless devices always desire such devices to be of small volume, light weight, and low cost. Thanks to the rapid advances in very large scale integration (VLSI) technology, this dream has become a reality in the past two decades. As technology grows rapidly, a mobile is not just a phone recently. The highly integration of circuits makes the mobile phone and the PDA (personal digital assistant) been combined into a single handset, which is called a smart phone. Also, the Internet carries various information resources and services, such as electronic mail, online chat, file transfer and file sharing, these attractive proprieties make wireless internet service becomes an important function that should be integrated into mobile devices. There are many ways for the user to connect to the internet. The traditional wireless local area network (WLAN) is a popular communication system for accessing the Internet. However, the reach of WiFi is very limited. WLAN connectivity is primarily constrained to hotspots, users need to find the access points and can only use it in certain rooms or areas. As the user get out of range of the hotspot, the signal will become very weak and the user may lose the connection. This disadvantage limits the mobility of wireless communication. Except for the widely used wireless local area network, third generation (3G) mobile telephony based on the High Speed Downlink Packet Access (HSDPA), which is part of the UMTS standards in 3G communications protocol, is another high speed wireless internet access service. It has become popular nowadays that people can get to the internet via cellular communication system. This technology gives the users the ability to access to the Internet wherever the signal is available from the cellular base station. However, the quality sometimes depends on the number of users simultaneously connected per cellular site. In addition to utilizing WLAN/3G dual-mode terminals to enhance efficiency of mobile number portability service, WiMAX (the Worldwide Interoperability for Microwave Access) is an emerging telecommunications technology that provides wireless data transmission in a variety of ways, ranging from point-to-point links to full mobile cellular-type access. WiMAX is similar to Wi-Fi but it can also permit usage at much greater distances. The bandwidth and range of WiMAX make it suitable for the applications like VoIP (Voice over Internet Protocol) or IPTV (Internet Protocol Television). Many people expect WiMAX to emerge as another technology that may be adopted for handset devices in

**1. Introduction** 

the near future.

**Design for Wireless Mobile Terminal** 

*2Department of Electric Engineering, National Ilan University,* 

*1Department of Electrical Engineering, National Chiao Tung University,* 

Yu-Jen Chi1 and Chien-Wen Chiu2

### **Hexa-Band Multi-Standard Planar Antenna Design for Wireless Mobile Terminal**

Yu-Jen Chi1 and Chien-Wen Chiu2

*1Department of Electrical Engineering, National Chiao Tung University, 2Department of Electric Engineering, National Ilan University, Taiwan* 

#### **1. Introduction**

Electronic devices such as mobile phones and laptop computers are parts of modern life. Users of portable wireless devices always desire such devices to be of small volume, light weight, and low cost. Thanks to the rapid advances in very large scale integration (VLSI) technology, this dream has become a reality in the past two decades. As technology grows rapidly, a mobile is not just a phone recently. The highly integration of circuits makes the mobile phone and the PDA (personal digital assistant) been combined into a single handset, which is called a smart phone. Also, the Internet carries various information resources and services, such as electronic mail, online chat, file transfer and file sharing, these attractive proprieties make wireless internet service becomes an important function that should be integrated into mobile devices. There are many ways for the user to connect to the internet. The traditional wireless local area network (WLAN) is a popular communication system for accessing the Internet. However, the reach of WiFi is very limited. WLAN connectivity is primarily constrained to hotspots, users need to find the access points and can only use it in certain rooms or areas. As the user get out of range of the hotspot, the signal will become very weak and the user may lose the connection. This disadvantage limits the mobility of wireless communication. Except for the widely used wireless local area network, third generation (3G) mobile telephony based on the High Speed Downlink Packet Access (HSDPA), which is part of the UMTS standards in 3G communications protocol, is another high speed wireless internet access service. It has become popular nowadays that people can get to the internet via cellular communication system. This technology gives the users the ability to access to the Internet wherever the signal is available from the cellular base station. However, the quality sometimes depends on the number of users simultaneously connected per cellular site. In addition to utilizing WLAN/3G dual-mode terminals to enhance efficiency of mobile number portability service, WiMAX (the Worldwide Interoperability for Microwave Access) is an emerging telecommunications technology that provides wireless data transmission in a variety of ways, ranging from point-to-point links to full mobile cellular-type access. WiMAX is similar to Wi-Fi but it can also permit usage at much greater distances. The bandwidth and range of WiMAX make it suitable for the applications like VoIP (Voice over Internet Protocol) or IPTV (Internet Protocol Television). Many people expect WiMAX to emerge as another technology that may be adopted for handset devices in the near future.

Hexa-Band Multi-Standard Planar Antenna Design for Wireless Mobile Terminal 5

bandwidth. This hepta-band antenna is designed for a mobile device and the parasitic element broadens the impedance bandwidth to about 45.5%. This antenna is extended to simultaneously operate in WLAN, WiMAX, and WWAN systems. It covers all cellular bands world-wide and all wireless network bands, such as the following communication standards: GSM/DCS/PCS/UMTS/WLAN/WiMAX/HIPERLAN2/IEEE 802.11. The antenna structure that measures only 50 mm x 12 mm x 0.5 mm can be easily fabricated by stamping from a metal plate. The following describes the details of the proposed antenna as

Parasitic Element 1

Ground Plane

(a)

(b) Fig. 1. The proposed antenna (a) Three-dimensional configuration of the proposed antenna

100mm

50mm

x z y

well as the experimental results.

**L**

Dual Band Main

(b) Plane view of the antenna structure.

Resonator

Parasitic Element 2

Shorting Strip

Feeding Point

The rapid progress in mobile communication requires that many functions and wireless communication systems be integrated into a mobile phone. When portability is taken into account, antenna that can be built in the phone device is desirable. This has led to a great demand for designing multiband antennas for handset devices. Among existing built-in or internal type scheme, the inverted-F (IFA) or planar inverted-F antenna (PIFA) are the most promising candidates. The linear inverted-F antenna, which is the original version of the PIFA, has been described by R. King in 1960 as a shunt-driven inverted-L antennatransmission line with open-end (king et al., 1960). The PIFA, which is constructed by replacing the linear radiator element of IFA with a planar radiator element, can also be evolved from a microstrip antenna. Taga first investigated PIFA's performance for 800MHz band portable unit radio in 1987 (Taga & Tsunekawa, 1987). He also wrote a chapter in his textbook to teach how to design a single band PIFA (Hirasawa & Haneishi, 1922). The PIFA or IFA are not only small in size but also have a broadband bandwidth. Since it is cheap and easy to fabricate, it has become very popular with mobile phone manufacturers. Many references concerning PIFA and its relatives were published in the decade.

In the past decade, researches for variation of the PIFA and multiband antenna grow rapidly like mushroom. Tri-band, quad-band, penta-band or hexa-band antenna can be found in many journals (Chiu & Lin, 2002; Guo et al., 2003, 2004; Ciais et al., 2004; Chen, 2007; Bancroft, 2005; Ali & Hayes, 2000; Soras et al., 2002; Nepa et al., 2005; Wong et al., 2005; Liu & Gaucher, 2004, 2007; Wang et al., 2007). For example, Chiu presented a tri-band PIFA for GSM800/DCS1800/PCS1900 in 2002 (Chiu & Lin, 2002) . Using two folded arms between the two plates, Guo at el. proposed a compact internal quad-band for covering GSM900/DCS1800/PCS1900 and ISM2450 bands (Guo, et al., 2003). By adding three quarter-wavelength parasitic elements to create new resonances, Ciais et al. presented a design of a compact quad-band PIFA for mobile phones (Ciais et al., 2004). In 2004, Guo & Tan proposed a new compact six-band but complicated internal antenna. His antenna is comprised of a main plate, a ground plane, a parasitic plate and a folded stub perpendicular to the two main plates (Guo & Tan, 2004).

In order to integrate all the wireless services into a mobile terminal and have an effective usage of the precious board space in the mobile device, multiband antenna that is designed to operate on several bands is necessary. However, designing a multiband antenna in a narrow space is a great challenge; a method that decrease the complexity of the antenna structure is also necessary to be investigated. Guo et. al. have recently designed quad-band antennas for mobile phones (Chiu & Lin, 2002; Nashaat et al., 2005; Karkkainen, 2005) and dual-band antennas for WLAN operations (Su & Chou, 2008). However, few of these antennas simultaneously cover the following communication standards: GSM (880-960 MHz), DCS (1710-1880 MHz), PCS (1850-1990 MHz), UMTS2100 (1920-2170 MHz), WLAN + Bluetooth (2400-2480 MHz), WiMAX (2500-2690 MHz), HiperLAN/2 in Europe (5150-5350 / 5470-5725 MHz) and IEEE 802.11a in the U.S. (5150-5350 / 5725-5825 MHz) (Liu & Gaucher, 2004, 2007; Wang et al., 2007; Rao & Geyi, 2009; Nguyen et al., 2009; Anguera et al., 2010; Kumar et al., 2010; Liu et al., 2010; Hsieh et al., 2009; Yu & Tarng, 2009; Hong et at., 2008; Guo et al., 2004; Li et al., 2010). This chapter proposes a planar multiband antenna that comprises a dual-band inverted-F resonator and two parasitic elements to cover all the communication standards mentioned above. One element is devoted to generating a dipole mode and another is helpful to excite a loop mode so as to broaden the impedance

The rapid progress in mobile communication requires that many functions and wireless communication systems be integrated into a mobile phone. When portability is taken into account, antenna that can be built in the phone device is desirable. This has led to a great demand for designing multiband antennas for handset devices. Among existing built-in or internal type scheme, the inverted-F (IFA) or planar inverted-F antenna (PIFA) are the most promising candidates. The linear inverted-F antenna, which is the original version of the PIFA, has been described by R. King in 1960 as a shunt-driven inverted-L antennatransmission line with open-end (king et al., 1960). The PIFA, which is constructed by replacing the linear radiator element of IFA with a planar radiator element, can also be evolved from a microstrip antenna. Taga first investigated PIFA's performance for 800MHz band portable unit radio in 1987 (Taga & Tsunekawa, 1987). He also wrote a chapter in his textbook to teach how to design a single band PIFA (Hirasawa & Haneishi, 1922). The PIFA or IFA are not only small in size but also have a broadband bandwidth. Since it is cheap and easy to fabricate, it has become very popular with mobile phone manufacturers. Many

In the past decade, researches for variation of the PIFA and multiband antenna grow rapidly like mushroom. Tri-band, quad-band, penta-band or hexa-band antenna can be found in many journals (Chiu & Lin, 2002; Guo et al., 2003, 2004; Ciais et al., 2004; Chen, 2007; Bancroft, 2005; Ali & Hayes, 2000; Soras et al., 2002; Nepa et al., 2005; Wong et al., 2005; Liu & Gaucher, 2004, 2007; Wang et al., 2007). For example, Chiu presented a tri-band PIFA for GSM800/DCS1800/PCS1900 in 2002 (Chiu & Lin, 2002) . Using two folded arms between the two plates, Guo at el. proposed a compact internal quad-band for covering GSM900/DCS1800/PCS1900 and ISM2450 bands (Guo, et al., 2003). By adding three quarter-wavelength parasitic elements to create new resonances, Ciais et al. presented a design of a compact quad-band PIFA for mobile phones (Ciais et al., 2004). In 2004, Guo & Tan proposed a new compact six-band but complicated internal antenna. His antenna is comprised of a main plate, a ground plane, a parasitic plate and a folded stub perpendicular

In order to integrate all the wireless services into a mobile terminal and have an effective usage of the precious board space in the mobile device, multiband antenna that is designed to operate on several bands is necessary. However, designing a multiband antenna in a narrow space is a great challenge; a method that decrease the complexity of the antenna structure is also necessary to be investigated. Guo et. al. have recently designed quad-band antennas for mobile phones (Chiu & Lin, 2002; Nashaat et al., 2005; Karkkainen, 2005) and dual-band antennas for WLAN operations (Su & Chou, 2008). However, few of these antennas simultaneously cover the following communication standards: GSM (880-960 MHz), DCS (1710-1880 MHz), PCS (1850-1990 MHz), UMTS2100 (1920-2170 MHz), WLAN + Bluetooth (2400-2480 MHz), WiMAX (2500-2690 MHz), HiperLAN/2 in Europe (5150-5350 / 5470-5725 MHz) and IEEE 802.11a in the U.S. (5150-5350 / 5725-5825 MHz) (Liu & Gaucher, 2004, 2007; Wang et al., 2007; Rao & Geyi, 2009; Nguyen et al., 2009; Anguera et al., 2010; Kumar et al., 2010; Liu et al., 2010; Hsieh et al., 2009; Yu & Tarng, 2009; Hong et at., 2008; Guo et al., 2004; Li et al., 2010). This chapter proposes a planar multiband antenna that comprises a dual-band inverted-F resonator and two parasitic elements to cover all the communication standards mentioned above. One element is devoted to generating a dipole mode and another is helpful to excite a loop mode so as to broaden the impedance

references concerning PIFA and its relatives were published in the decade.

to the two main plates (Guo & Tan, 2004).

bandwidth. This hepta-band antenna is designed for a mobile device and the parasitic element broadens the impedance bandwidth to about 45.5%. This antenna is extended to simultaneously operate in WLAN, WiMAX, and WWAN systems. It covers all cellular bands world-wide and all wireless network bands, such as the following communication standards: GSM/DCS/PCS/UMTS/WLAN/WiMAX/HIPERLAN2/IEEE 802.11. The antenna structure that measures only 50 mm x 12 mm x 0.5 mm can be easily fabricated by stamping from a metal plate. The following describes the details of the proposed antenna as well as the experimental results.

Fig. 1. The proposed antenna (a) Three-dimensional configuration of the proposed antenna (b) Plane view of the antenna structure.

Hexa-Band Multi-Standard Planar Antenna Design for Wireless Mobile Terminal 7

0.5 1 1.5 Frequency (GHz)

Ls

Fig. 4 shows another kind of inverted-F antenna while the shorting pin is moved to the bottom for size reduction. The mechanism of this alternative is the same as the previous one,

The dual band inverted-F antenna can be simply accomplished by creating two resonant paths of the antenna element. As shown in Fig. 5, the dual-band main resonator consists of two branches (A and B). The length of the longer branch (B) is about 83 mm (9 + 44.5 + 6 + 23.5 mm) which is one-quarter of the wavelength at 900MHz. The lower resonant mode for GSM operation can be excited on this resonator. On the other hand, branch (A) in the middle creates a shorter path of 42 mm, which is about a quarter of wavelength at 1800 MHz. As a result, the resonant mode for DCS operation can be excited. Simulation result of the dual band antenna is shown in Fig. 6. The input impedance can be adjusted by changing the

but the input impedance is matched by adjusting the length of the shorting strip LS.

Fig. 3. Antenna height influences on the impedance bandwidth for a simple IFA.

H = 5mm H = 6mm H = 7mm H = 8mm H = 9mm H = 10mm



H

Fig. 4. A variation of typical inverted-F antenna.


Return Loss (dB)

│S11│ (dB)

0

10

#### **2 Antenna design**

#### **2.1 Design of a dual-band antenna**

Modern mobile terminals require small and thin design, therefore, planar inverted-F antenna, which requires a spacing of about 7 mm ~ 12 mm between the antenna and the substrate to achieve the sufficient operating bandwidth, is not suitable to be integrated with the present thin mobile terminals although it is popular and widely used. Fig. 1(a) shows a three dimensional view of the proposed design. The antenna, which is mounted on the top edge of the printed circuit board (PCB), is fed by a 50 Ω coaxial cable. The antenna is coplanar with the system ground of the PCB. The dielectric constant of the PCB used here is 4.4 and the thickness is 1.58 mm. As shown in Fig. 1(b), this radiating structure measures 50 mm × 12 mm × 1.5 mm and can be extended to a single metallic plate. It is basically an inverted-F antenna in which the quarter-wavelength characteristic is obtained thanks to a short-circuited metallic strip. As indicated in Fig. 1(b), this design comprises a direct-feed dual band main resonator with two branches (A) and (B), and two parasitic elements (C) and (D) excited by electromagnetic coupling, to achieve multiband operation.

Shown in Fig. 2 is a typical configuration of an inverted-F antenna. It can be fed by a minicoaxial cable which is connected to the RF module. Here, H is the height of the radiator above the ground plane, LF is the horizontal length from the feed point to the open end of the antenna, and LB is the horizontal length from the feed point to the closed end of the antenna. This antenna is a quarter-wavelength radiator with one short end and one open end. The resonant frequency can be easily calculated by the formula:

$$f = \frac{c}{4(H + L\_B + L\_F)}$$

the where *c* is the speed of light. The resonant frequency can be adjusted by changing the value LF, and the distance LB between the feed point and shorting strip can be used to adjust the input impedance. The height H of the antenna is closely related to the impedance bandwidth where the Q factor can be reduced by increasing the antenna height to broaden the bandwidth and vice versa. Variations of IFA Antenna height cause some effects on bandwidth. Fig. 3 shows the simulation results with different antenna height H. It is found that increasing the height will increase the impedance bandwidth.

Fig. 2. A typical inverted-F Antenna.

Modern mobile terminals require small and thin design, therefore, planar inverted-F antenna, which requires a spacing of about 7 mm ~ 12 mm between the antenna and the substrate to achieve the sufficient operating bandwidth, is not suitable to be integrated with the present thin mobile terminals although it is popular and widely used. Fig. 1(a) shows a three dimensional view of the proposed design. The antenna, which is mounted on the top edge of the printed circuit board (PCB), is fed by a 50 Ω coaxial cable. The antenna is coplanar with the system ground of the PCB. The dielectric constant of the PCB used here is 4.4 and the thickness is 1.58 mm. As shown in Fig. 1(b), this radiating structure measures 50 mm × 12 mm × 1.5 mm and can be extended to a single metallic plate. It is basically an inverted-F antenna in which the quarter-wavelength characteristic is obtained thanks to a short-circuited metallic strip. As indicated in Fig. 1(b), this design comprises a direct-feed dual band main resonator with two branches (A) and (B), and two parasitic elements (C)

and (D) excited by electromagnetic coupling, to achieve multiband operation.

end. The resonant frequency can be easily calculated by the formula:

that increasing the height will increase the impedance bandwidth.

Fig. 2. A typical inverted-F Antenna.

Shown in Fig. 2 is a typical configuration of an inverted-F antenna. It can be fed by a minicoaxial cable which is connected to the RF module. Here, H is the height of the radiator above the ground plane, LF is the horizontal length from the feed point to the open end of the antenna, and LB is the horizontal length from the feed point to the closed end of the antenna. This antenna is a quarter-wavelength radiator with one short end and one open

> 4( ) *B F <sup>c</sup> <sup>f</sup> HL L*

the where *c* is the speed of light. The resonant frequency can be adjusted by changing the value LF, and the distance LB between the feed point and shorting strip can be used to adjust the input impedance. The height H of the antenna is closely related to the impedance bandwidth where the Q factor can be reduced by increasing the antenna height to broaden the bandwidth and vice versa. Variations of IFA Antenna height cause some effects on bandwidth. Fig. 3 shows the simulation results with different antenna height H. It is found

**2 Antenna design** 

**2.1 Design of a dual-band antenna** 

Fig. 3. Antenna height influences on the impedance bandwidth for a simple IFA.

Fig. 4. A variation of typical inverted-F antenna.

Fig. 4 shows another kind of inverted-F antenna while the shorting pin is moved to the bottom for size reduction. The mechanism of this alternative is the same as the previous one, but the input impedance is matched by adjusting the length of the shorting strip LS.

The dual band inverted-F antenna can be simply accomplished by creating two resonant paths of the antenna element. As shown in Fig. 5, the dual-band main resonator consists of two branches (A and B). The length of the longer branch (B) is about 83 mm (9 + 44.5 + 6 + 23.5 mm) which is one-quarter of the wavelength at 900MHz. The lower resonant mode for GSM operation can be excited on this resonator. On the other hand, branch (A) in the middle creates a shorter path of 42 mm, which is about a quarter of wavelength at 1800 MHz. As a result, the resonant mode for DCS operation can be excited. Simulation result of the dual band antenna is shown in Fig. 6. The input impedance can be adjusted by changing the

Hexa-Band Multi-Standard Planar Antenna Design for Wireless Mobile Terminal 9

resonator C, and the main resonators A and B. Fig. 7 shows the surface current distributions on the resonators and the ground plane. Finding show that part of the dual band resonator and the parasitic element form a dipole antenna. From point a, through point b, c, and d, then to point f in Fig. 7, the total length (39 mm + 3 mm + 9 mm + 19 mm = 70 mm) is closed to 0.5 wavelength at 2250MHz (67 mm). This allows the antenna to generate an additional

> **0.5 1 1.5 2 2.5 3 Frequency (GHz)**

Fig. 8. Parameter study with different length of the parasitic resonator.

 **L = 13mm L = 15mm L = 17mm L = 19mm**

0.5-wavelength resonant mode at 2250 MHz to cover the desired operation bands.

Fig. 7. Victor surface current distribution at 2.25 GHz.

**-40**

**-35**

**-30**

**-25**

**-20**

**│S11│ (dB)** 

**-15**

**-10**

**-5**

**0**

**5**

length of the shorting strip *Ls* . In this case, *Ls* is selected to be 22.5 mm to have the widest bandwidth at both lower and upper band.

Fig. 5. A dual band inverted-F main resonator.

Fig. 6. Parameter study with different value of Ls.

#### **2.2 Bandwidth enhanced by a parasitic element**

Creating multiple resonant paths of the inverted-F antenna is helpful to generate multiple resonances. However, the coupling between each resonant path makes it difficult to match the antenna at each frequency band. To cover the wide bandwidth from 1900 MHz to 2700 MHz, this work introduces a parasitic resonator C near the main driven resonator. This parasitic element is excited by electromagnetic coupling from the main dual band resonator. Thus, a dipole-like antenna that resonates at 2250 MHz is formed by both the introduced

length of the shorting strip *Ls* . In this case, *Ls* is selected to be 22.5 mm to have the widest

**0.5 1 1.5 2 Frequency (GHz)**

Creating multiple resonant paths of the inverted-F antenna is helpful to generate multiple resonances. However, the coupling between each resonant path makes it difficult to match the antenna at each frequency band. To cover the wide bandwidth from 1900 MHz to 2700 MHz, this work introduces a parasitic resonator C near the main driven resonator. This parasitic element is excited by electromagnetic coupling from the main dual band resonator. Thus, a dipole-like antenna that resonates at 2250 MHz is formed by both the introduced

**Ls = 34.5 mm Ls = 22.5 mm Ls = 10.5 mm**

bandwidth at both lower and upper band.

Fig. 5. A dual band inverted-F main resonator.

**-25**

Fig. 6. Parameter study with different value of Ls.

**2.2 Bandwidth enhanced by a parasitic element** 

**-20**

**-15**

**-10**

**│S11│ (dB)** 

**-5**

**0**

**5**

resonator C, and the main resonators A and B. Fig. 7 shows the surface current distributions on the resonators and the ground plane. Finding show that part of the dual band resonator and the parasitic element form a dipole antenna. From point a, through point b, c, and d, then to point f in Fig. 7, the total length (39 mm + 3 mm + 9 mm + 19 mm = 70 mm) is closed to 0.5 wavelength at 2250MHz (67 mm). This allows the antenna to generate an additional 0.5-wavelength resonant mode at 2250 MHz to cover the desired operation bands.

Fig. 7. Victor surface current distribution at 2.25 GHz.

Fig. 8. Parameter study with different length of the parasitic resonator.

Hexa-Band Multi-Standard Planar Antenna Design for Wireless Mobile Terminal 11

25.5 mm + 1 mm) is roughly equal to a wavelength of the resonant frequency 5.59 GHz (53.67 mm). Fig. 10 (b) shows surface current distributions at the resonating frequency 5.59 GHz, The vector current distribution shown in Fig. 11(b) demonstrates that one-wavelength

(a)

(b)

Fig. 10. Surface current distribution at (a) 5.20 and (b) 5.59 GHz.

loop mode is excited on the resonator E.

To demonstrate the effect of the parasitic element covering from 1900 MHz to 2700 MHz, Fig. 8 shows the parameter study of the proposed antennas with different length of the parasitic element. By Investigating the Smith chart shown in Fig. 9, it is evident that the input impedance is closer to 50 Ω as length L increases, because the longer the parasitic element, the more the loaded capacitance (Chi, 2009). The narrow gap between the main resonator and the parasitic element C introduces a proper capacitance to compensate for possible inductance contributed from the dual-band main resonator. Increasing capacitance neutralizes the effect due to inductance of the strip. Therefore, the capacitive coupled parasitic element creates a new resonant mode but does not change the original two resonant modes at 900 MHz and 1800 MHz. The length of the parasitic element is selected to be 19 mm to have the return loss better than 6 dB in the band of operation. The achieve bandwidth of the parasitic element is about 34.78 %, covering from 1900 MHz to 2700 MHz, which is enough for WLAN, WMAN, and WWAN operations.

Fig. 9. Parametric study – Smith Chart.

#### **2.3 Create resonances at the U-NII band**

So far, a hexa-band Inverted-F antenna has been designed, except IEEE 802.11a or HYPERLAN/2. The current research will include the U-NII (Unlicensed National Information Infrastructure) band in this design by a tuning parasitic resonator D, as Fig. 1(b) shows. First, the third harmonics of the resonating frequency in the second band (1.72 GHz) is about 5.20 GHz. This mode which contributes to the U-NII band is also excited. The surface current distribution on the resonator A in Fig. 10(a) demonstrates that the 1.5 wavelength mode generates at the resonating frequency. The vector current distribution is shown in Fig. 11(a). Second, the loop resonator E in Fig. 1(b) is designed as a onewavelength rectangular loop antenna. The perimeter of the loop antenna (25.5 mm + 1 mm +

To demonstrate the effect of the parasitic element covering from 1900 MHz to 2700 MHz, Fig. 8 shows the parameter study of the proposed antennas with different length of the parasitic element. By Investigating the Smith chart shown in Fig. 9, it is evident that the input impedance is closer to 50 Ω as length L increases, because the longer the parasitic element, the more the loaded capacitance (Chi, 2009). The narrow gap between the main resonator and the parasitic element C introduces a proper capacitance to compensate for possible inductance contributed from the dual-band main resonator. Increasing capacitance neutralizes the effect due to inductance of the strip. Therefore, the capacitive coupled parasitic element creates a new resonant mode but does not change the original two resonant modes at 900 MHz and 1800 MHz. The length of the parasitic element is selected to be 19 mm to have the return loss better than 6 dB in the band of operation. The achieve bandwidth of the parasitic element is about 34.78 %, covering from 1900 MHz to 2700 MHz,

So far, a hexa-band Inverted-F antenna has been designed, except IEEE 802.11a or HYPERLAN/2. The current research will include the U-NII (Unlicensed National Information Infrastructure) band in this design by a tuning parasitic resonator D, as Fig. 1(b) shows. First, the third harmonics of the resonating frequency in the second band (1.72 GHz) is about 5.20 GHz. This mode which contributes to the U-NII band is also excited. The surface current distribution on the resonator A in Fig. 10(a) demonstrates that the 1.5 wavelength mode generates at the resonating frequency. The vector current distribution is shown in Fig. 11(a). Second, the loop resonator E in Fig. 1(b) is designed as a onewavelength rectangular loop antenna. The perimeter of the loop antenna (25.5 mm + 1 mm +

which is enough for WLAN, WMAN, and WWAN operations.

Fig. 9. Parametric study – Smith Chart.

**2.3 Create resonances at the U-NII band** 

25.5 mm + 1 mm) is roughly equal to a wavelength of the resonant frequency 5.59 GHz (53.67 mm). Fig. 10 (b) shows surface current distributions at the resonating frequency 5.59 GHz, The vector current distribution shown in Fig. 11(b) demonstrates that one-wavelength loop mode is excited on the resonator E.

Fig. 10. Surface current distribution at (a) 5.20 and (b) 5.59 GHz.

Hexa-Band Multi-Standard Planar Antenna Design for Wireless Mobile Terminal 13

measured data and simulated results. The antenna covers all cellular bands used worldwide is evident. The achieved bandwidths with return loss better than 6 dB are 80 MHz (880–960 MHz) in the GSM band, 1000 MHz (1700–2700 MHz) in the DCS/PCS/UMTS/WiFi /WiMAX band and 1270 MHz (4820–6090 MHz) in the 5 GHz U-NII band. When ground plane length varies from 80 mm to 120 mm, frequency shifting is

(a) (b)

**-35**

Fig. 13. Measured and simulated results of the proposed antenna.

**-30**

**-25**

**-20**

**-15**

**Return Loss (dB)**

**│S11│ (dB)** 

**-10**

**-5**

**0**

**5**

Fig. 12. Photography of the fabricated antenna (a) top view, (b) side view.

**0.5 1.5 2.5 3.5 4.5 5.5 6.5 Frequency (GHz)**

**Simulated (HFSS)**


**Measured**

slight (Chi, 2009).

Fig. 11. Victor current distributions at higher U-NII bands: (a) 5.20 and (b) 5.59 GHz.

Finally, this work applies another technique to tune the higher order resonances for the U-NII band. The quarter wavelength resonating at 6.0 GHz is only about 12.5 mm. A short resonator D with a length of 10.5 mm, as Fig. 1(b) shows, is introduced to the short-circuited pin of the main resonator to form an inverted L-shape parasitic element. The capacitive coupling between the strip and the chassis increases its electrical length since the radiating strip is only 1 mm above the ground plane. Adding this parasitic element improves resonance performance at the U-NII band.

#### **3. Results and discussion**

This study constructs and tests the proposed antenna based on the design dimensions shown in Fig. 1(b). The test structure was shown in Fig. 12 and the measurement of scattering parameters was performed by an Agilent E5071B network analyzer. Fig. 13 shows the measured and simulated return loss where the solid red line is the measured result and the dotted blue line is the simulated one. Findings show good agreement between the

(a)

(b)

Finally, this work applies another technique to tune the higher order resonances for the U-NII band. The quarter wavelength resonating at 6.0 GHz is only about 12.5 mm. A short resonator D with a length of 10.5 mm, as Fig. 1(b) shows, is introduced to the short-circuited pin of the main resonator to form an inverted L-shape parasitic element. The capacitive coupling between the strip and the chassis increases its electrical length since the radiating strip is only 1 mm above the ground plane. Adding this parasitic element improves

This study constructs and tests the proposed antenna based on the design dimensions shown in Fig. 1(b). The test structure was shown in Fig. 12 and the measurement of scattering parameters was performed by an Agilent E5071B network analyzer. Fig. 13 shows the measured and simulated return loss where the solid red line is the measured result and the dotted blue line is the simulated one. Findings show good agreement between the

Fig. 11. Victor current distributions at higher U-NII bands: (a) 5.20 and (b) 5.59 GHz.

resonance performance at the U-NII band.

**3. Results and discussion** 

measured data and simulated results. The antenna covers all cellular bands used worldwide is evident. The achieved bandwidths with return loss better than 6 dB are 80 MHz (880–960 MHz) in the GSM band, 1000 MHz (1700–2700 MHz) in the DCS/PCS/UMTS/WiFi /WiMAX band and 1270 MHz (4820–6090 MHz) in the 5 GHz U-NII band. When ground plane length varies from 80 mm to 120 mm, frequency shifting is slight (Chi, 2009).

(a) (b)

Fig. 12. Photography of the fabricated antenna (a) top view, (b) side view.

Fig. 13. Measured and simulated results of the proposed antenna.

Hexa-Band Multi-Standard Planar Antenna Design for Wireless Mobile Terminal 15

0

180

(a) 925 MHz

0

180

(b) 2170 MHz

0

180

(c) 2650 MHz

x-y plane x-z plane y-z plane

x-y plane x-z plane y-z plane

x-y plane x-z plane y-z plane

90

90

90

270

**-35 -25 -15 -5 5**

270

**-35 -25 -15 -5 5**

270

0

180

0

180

0

180

90

90

90

**-35 -25 -15 -5 5**

**-35 -25 -15 -5 5**

**-35 -25 -15 -5 5**

**-35 -25 -15 -5 5**

0

180

0

180

0

180

270

270

270

90

90

90

270

**-35 -25 -15 -5 5**

270

**-35 -25 -15 -5 5**

270

**-35 -25 -15 -5 5**

This study performed radiation-pattern and gain measurement in the anechoic chambers of SGS Ltd. Taiwan, as shown in Fig. 14. Fig. 15 shows the measured and simulated radiation patterns at the xy-cut, xz-cut, and yz-cut. The measured radiation patterns show a good match to the simulation results except at 925MHz. In the small antenna measurement, the patterns are easily affected by the feeding RF cable in the GSM band (Chen et al., 2005). This work finds that the dual-polarization radiation-patterns have very suitable characteristics for portable devices. For the radiation shown in Fig. 14(a), more energy for Eθ is radiated in the lower band as compared to Eφ. The Eφ field has some dips at 900 MHz on the xz-plane or 1800 MHz on the xy-plane. This is probably due to current cancellation on the strips and the ground plane.

Fig. 14. Radiation Pattern measurement in a 3D anechoic chamber.

Findings also show a dipole-like pattern at the frequency 2170 MHz. Radiation patterns shown in Fig. 15(b) confirm this deduction. The radiation pattern of this mode is similar to a small dipole oriented in the y–axis leading to a directional pattern in the E-plane (xy-plane, blue line) and omni-directional pattern in the H-plane (xz-plane, blue line), as Fig. 15(b), shows respectively. The resonators C and B at 2170 MHz have strong current distributions along the z-direction which also contribute to radiation fields. The radiation pattern of this current distribution is due to a small dipole oriented in the z–axis leading to a bidirectional pattern in the E-plane (xz-plane, red line) and omni-directional pattern in the H-plane (xyplane, red line), as Fig. 15(b), shows respectively. Findings also show an asymmetric radiation pattern at the U-NII band (5-6 GHz) and some variation and nulls, since different modes are excited in this U-NII band.

This study performed radiation-pattern and gain measurement in the anechoic chambers of SGS Ltd. Taiwan, as shown in Fig. 14. Fig. 15 shows the measured and simulated radiation patterns at the xy-cut, xz-cut, and yz-cut. The measured radiation patterns show a good match to the simulation results except at 925MHz. In the small antenna measurement, the patterns are easily affected by the feeding RF cable in the GSM band (Chen et al., 2005). This work finds that the dual-polarization radiation-patterns have very suitable characteristics for portable devices. For the radiation shown in Fig. 14(a), more energy for Eθ is radiated in the lower band as compared to Eφ. The Eφ field has some dips at 900 MHz on the xz-plane or 1800 MHz on the xy-plane. This is probably due to current cancellation on the strips and

Fig. 14. Radiation Pattern measurement in a 3D anechoic chamber.

modes are excited in this U-NII band.

Findings also show a dipole-like pattern at the frequency 2170 MHz. Radiation patterns shown in Fig. 15(b) confirm this deduction. The radiation pattern of this mode is similar to a small dipole oriented in the y–axis leading to a directional pattern in the E-plane (xy-plane, blue line) and omni-directional pattern in the H-plane (xz-plane, blue line), as Fig. 15(b), shows respectively. The resonators C and B at 2170 MHz have strong current distributions along the z-direction which also contribute to radiation fields. The radiation pattern of this current distribution is due to a small dipole oriented in the z–axis leading to a bidirectional pattern in the E-plane (xz-plane, red line) and omni-directional pattern in the H-plane (xyplane, red line), as Fig. 15(b), shows respectively. Findings also show an asymmetric radiation pattern at the U-NII band (5-6 GHz) and some variation and nulls, since different

the ground plane.

(a) 925 MHz

(b) 2170 MHz

Hexa-Band Multi-Standard Planar Antenna Design for Wireless Mobile Terminal 17

dimensional average gain and radiation efficiency for all the operation bands, showing that

This chapter reported a down-sized multiband inverted-F antenna to integrate the 3.5G and WLAN/WiMAX antenna systems. It is comprised of a dual-band antenna with one feed point and two parasitic elements to cover many mobile communication systems including GSM900 /DCS /PCS /UMTS /WLAN/ WiMAX /HiperLAN2 /IEEE802.11a. Measured parameters including return loss, radiation patterns, three-dimensional peak gain and average gain as well as radiation efficiency were presented to validate the proposed design. Since this antenna can be formed by a single plate, it is both low cost and easy to fabricate,

C. Soras, M. Karaboikis, and G. T. V. Makios, "Analysis and design of an inverted-F antenna

C. W. Chiu and F. L. Lin, "Compact dual-band PIFA with multi-resonators," Electronics

C.-L. Liu, Y.-F. Lin, C.-M. Liang, S.-C. Pan, and H.-M. Chen, "Miniature Internal Penta-Band

D. Liu and B. Gaucher, "A new multiband antenna for WLAN/Cellular application," Vehicular Technology Conference, vol. 1, 60th, pp. 243 - 246, Sept. 2004. D. Liu and B. Gaucher, "A quadband antenna for laptop application," International

D.M. Nashaat, H. A. Elsadek, and H. Ghali, "Single feed compact quad -band PIFA antenna

H.-W. Hsieh, Y.-C. Lee, K.-K. Tiong, and J.-S. Sun, "Design of A Multiband Antenna for Mobile Handset Operations," IEEE Antennas Wireless Propag. Lett., vol. 8, 2009. J. Anguera, I. Sanz, J. Mumbrú, and C. Puente, "Multiband Handset Antenna with A Parallel

K. Hirasawa and M. Haneishi, "Analysis, design and measurement of small and low profile

K.-L. Wong, L.-C. Chou, and C.-M. Su, "Dual-band flat-plate antenna with a shorted

M. Ali and G. J. Hayes, "Analysis of intergated inverted-F antennas for bluetooth applications," IEEE International symposium on antenna and propagation, 2000. M. K. Karkkainen, "Meandered multiband PIFA with coplanar parasitic patches," IEEE

Workshop on Antenna Technology, pp. 128-131, March 2007.

antennas," ch.5, Norwood, MA, Artech House, 1922.

Propagation, vol. 53, no. 1, pp. 539-544, January 2005.

Microw. Wireless Compon. Lett., vol.15, pp. 630-632, Oct. 2005.

printed on a PCMCIA card for the 2.4 GHz ISM band," IEEE Antenna's and

Monopole Antenna for Mobile Phones," IEEE Trans. Antennas Propag., vol. 58, no.

for wireless communication applications," IEEE Trans. Antennas Propagat., vol. 53,

Excitation of PIFA and Slot Radiators," IEEE Trans. Antennas Propag., vol. 58, no.

parasitic element for laptop applications," IEEE Transactions on Antennas and

all radiation efficiencies are over 50 percent, meeting the specification requirement.

making it suitable for any palm-sized mobile device applications.

propagation magazine, vol. 44, no. 1, February 2002.

Letters, vol. 38, pp. 538-540, June 2002.

No. 8, pp. 2631-2635, Aug. 2005.

**4. Summary** 

**5. References** 

3, March 2010.

2, February 2010.

Fig. 15. Measured and simulated radiation patterns in three cuts (a) 925 MHz (b) 2170 MHz (c) 2650 MHz (d) 5775 MHz.


Table 1. Measured three-dimensional peak gain, average gain, and radiation efficiency.

By using the commercial electromagnetic simulation software HFSS, this research carries out simulations for the theoretical gains to investigate antenna performance and compare it with the measured results (Chi, 2009). Good agreement confirms that the measured data are accurate. The two-dimensional average gain is determined from pattern measurements made in the horizontal (azimuth) plane for both polarizations of the electric field. The results are then averaged over azimuth angles and normalized with respect to an ideal isotropic radiator (Chen, 2007). Finally, Table 1 lists the measured peak gain, twodimensional average gain and radiation efficiency for all the operation bands, showing that all radiation efficiencies are over 50 percent, meeting the specification requirement.

#### **4. Summary**

16 Advanced Transmission Techniques in WiMAX

0

180

(d) 5775 MHz

Fig. 15. Measured and simulated radiation patterns in three cuts (a) 925 MHz (b) 2170 MHz

Frequency (MHz) 925 1710 1795 1920 1990 Peak Gain (dBi) -0.25 2.4 2.05 1.39 1.63 Average Gain (dBi) -1.96 1.10 -0.63 -0.01 -0.51

Table 1. Measured three-dimensional peak gain, average gain, and radiation efficiency.

By using the commercial electromagnetic simulation software HFSS, this research carries out simulations for the theoretical gains to investigate antenna performance and compare it with the measured results (Chi, 2009). Good agreement confirms that the measured data are accurate. The two-dimensional average gain is determined from pattern measurements made in the horizontal (azimuth) plane for both polarizations of the electric field. The results are then averaged over azimuth angles and normalized with respect to an ideal isotropic radiator (Chen, 2007). Finally, Table 1 lists the measured peak gain, two-

Efficiency 51.42% 61.94% 64.85% 70.35% 78.80% Frequency 2170 2420 2650 5250 5800 Peak Gain 2.95 2.5 2.48 6.91 8.35 Average Gain 1.10 1.15 0.58 -0.31 -1.99 Efficiency 90.11% 86.83% 71.42% 70.24% 71.80%

x-y plane x-z plane y-z plane

90

**Measured E-theta Measured E-phi Simulated E-theta Simulated E-phi**

270

0

180

90

**-35 -25 -15 -5 5**

**-35 -25 -15 -5 5**

0

180

(c) 2650 MHz (d) 5775 MHz.

270

90

<sup>x</sup> <sup>y</sup>

z

270

**-35 -25 -15 -5 5**

This chapter reported a down-sized multiband inverted-F antenna to integrate the 3.5G and WLAN/WiMAX antenna systems. It is comprised of a dual-band antenna with one feed point and two parasitic elements to cover many mobile communication systems including GSM900 /DCS /PCS /UMTS /WLAN/ WiMAX /HiperLAN2 /IEEE802.11a. Measured parameters including return loss, radiation patterns, three-dimensional peak gain and average gain as well as radiation efficiency were presented to validate the proposed design. Since this antenna can be formed by a single plate, it is both low cost and easy to fabricate, making it suitable for any palm-sized mobile device applications.

#### **5. References**


**2** 

*Thailand* 

**CPW-Fed Antennas for WiFi and WiMAX** 

*Wireless Communication Research Group (WCRG), Electrical Engineering,* 

*Faculty of Engineering, King Mongkut's University of Technology North Bangkok,* 

Recently, several researchers have devoted large efforts to develop antennas that satisfy the demands of the wireless communication industry for improving performances, especially in term of multiband operations and miniaturization. As a matter of fact, the design and development of a single antenna working in two or more frequency bands, such as in wireless local area network (WLAN) or WiFi and worldwide interoperability for microwave access (WiMAX) is generally not an easy task. The IEEE 802.11 WLAN standard allocates the license-free spectrum of 2.4 GHz (2.40-2.48 GHz), 5.2 GHz (5.15-5.35 GHz) and 5.8 GHz (5.725-5.825 GHz). WiMAX, based on the IEEE 802.16 standard, has been evaluated by companies for last mile connectivity, which can reach a theoretical up to 30 mile radius coverage. The WiMAX forum has published three licenses spectrum profiles, namely the 2.3 (2.3-2.4 GHz), 2.5 GHz (2.495-2.69 GHz) and 3.5 GHz (3.5-3.6 GHz) varying country to country. Many people expect WiMAX to emerge as another technology especially WiFi that may be adopted for handset devices and base station in the near future. The eleven

Consequently, the research and manufacturing of both indoor and outdoor transmission equipment and devices fulfilling the requirements of these WiFi and WiMAX standards have increased since the idea took place in the technical and industrial community. An antenna serves as one of the critical component in any wireless communication system. As mentioned above, the design and development of a single antenna working in wideband or more frequency bands, called multiband antenna, is generally not an easy task. To answer these challenges, many antennas with wideband and/or multiband performances have been published in open literatures. The popular antenna for such applications is microstrip antenna (MSA) where several designs of multiband MSAs have been reported. Another important candidate, which may complete favorably with microstrip, is coplanar waveguide (CPW). Antennas using CPW-fed line also have many attractive features including lowradiation loss, less dispersion, easy integration for monolithic microwave circuits (MMICs) and a simple configuration with single metallic layer, since no backside processing is required for integration of devices. Therefore, the designs of CPW-fed antennas have recently become more and more attractive. One of the main issues with CPW-fed antennas is to provide an easy impedance matching to the CPW-fed line. In order to obtain multiband and broadband operations, several techniques have been reported in the literatures based on CPW-fed slot antennas (Chaimool et al., 2004, 2005, 2008; Sari-Kha et al., 2006; Jirasakulporn,

standardized WiFi and WiMAX operating bands are listed in Table I.

**1. Introduction** 

Sarawuth Chaimool and Prayoot Akkaraekthalin


### **CPW-Fed Antennas for WiFi and WiMAX**

Sarawuth Chaimool and Prayoot Akkaraekthalin

*Wireless Communication Research Group (WCRG), Electrical Engineering, Faculty of Engineering, King Mongkut's University of Technology North Bangkok, Thailand* 

#### **1. Introduction**

18 Advanced Transmission Techniques in WiMAX

P. Ciais, R. Staraj, G. Kossiavas, and C. Luxey, "Design of an internal quad-band antenna for

P. Kumar.m, S. Kumar, R. Jyoti, V. Reddy, and P. Rao1, "Novel Structural Design for

P.Nepa, G. Manara, A. A. Serra, and G. Nenna, "Multiband PIFA for WLAN mobile terminals," IEEE antenna and wireless propagation letters, vol. 4, 2005. Q. Rao and W. Geyi, "Compact Multiband Antenna for Handheld Devices," IEEE Trans.

R. Bancroft, "Development and integration of a commercially viable 802.11a/b/g HiperLan/

R. King, C. W. Harisson, and D. H. Denton, "Transmission-line missile antenna," IRE Trans.

S. Hong, W. Kim, H. Park, S. Kahng, and J. Choi, "Design of An Internal Multiresonant

S.W. Su and J.H. Chou, "Internal 3G and WLAN/WiMAX antennas integrated in palm-sized mobile devices," Microw. Opt. Technol. Lett., vol. 50, no. 1, pp. 29-31, Jan. 2008. T. K. Nguyen, B. Kim, H. Choo, and I. Park, "Multiband dual Spiral Stripline-Loaded Monopole Antenna," IEEE Antennas Wireless Propag. Lett., vol. 8, 2009. T. Taga and K. Tsunekawa, "Performance analysis pf a built-in planar inverted-F antenna for

W. X. Li, X. Liu, and S. Li, "Design of A Broadband and Multiband Planar Inverted-F

X. Wang, W. Chen, and Z. Feng, "Multiband antenna with parasitic branches for laptop applications," Electronics letters, vol. 43, no. 19, 13th, September 2007. Y. J., Chi, "Design of internal multiband antennas for portable devices," Master Thesis,

Y.-C. Yu and J.-H. Tarng, "A Novel Modified Multiband Planar Inverted-F Antenna," IEEE

Y.-X. Guo and H. S. Tan, "New compact six-band internal antenna," IEEE antenna and

Y.-X. Guo, I. Ang, and M. Y. W. Chia, "Compact internal multiband antennas for mobile handsets," IEEE antenna and wireless propogation letters, vol. 2, 2003. Y.-X. Guo, M. Y. W. Chia, and Z. N. Chen, "Miniature Built-In Multiband Antennas for Mobile Handsets," IEEE Trans. Antennas Propag., vol. 52, no. 8, August 2004.

Z. N. Chen, N. Yang, Y. X. Guo, and M. Y. W. Chia, "An investigation into measurement of

Zhi Ning Chen, "Antennas for Portable Devices," John Wiley & Sons, Inc. 2007, ch.4, pp.115-116.

handset antennas," IEEE. Trans. Instrum. Meas., vol. 54, no.3, pp. 1100–1110, June 2005.

Z. N. Chen, Antennas for Portable Devices, pp.125-126, John Wiley & Sons, Inc. 2007.

Antenna Technology (iWAT), 1-3 March 2010.

Antennas Propag., vol. 57, no. 10, October 2009.

Antenna Propagation, vol. 8, no. 1, pp. 88-90, 1960.

Trans. Antennas Propag., vol. 56, no. 5, May 2008.

communications, vol. SAC-5, no. 5, June 1987.

Antennas Wireless Propag. Lett., vol. 8, 2009.

wireless propagation letters, vol. 3, 2004.

Computing, vol. 2, 12-14 April 2010.

National Ilan University, June 2009

International Symposium, vol. 4A, pp. 231- 234, July 2005.

April 2004.

mobile phones," IEEE Microwave and wireless components letters, vol. 14, no. 4,

Compact and Broadband Patch Antenna," 2010 International Workshop on

WLAN antenna into laptop computers," Antennas and Propagation Society

Monopole Antenna for GSM900/DCS1800/US-PCS/S-DMB Operation," IEEE

800MHz and portable radio units," IEEE Trans. on selected areas in

Antenna," 2010 International Conference on Communications and Mobile

Recently, several researchers have devoted large efforts to develop antennas that satisfy the demands of the wireless communication industry for improving performances, especially in term of multiband operations and miniaturization. As a matter of fact, the design and development of a single antenna working in two or more frequency bands, such as in wireless local area network (WLAN) or WiFi and worldwide interoperability for microwave access (WiMAX) is generally not an easy task. The IEEE 802.11 WLAN standard allocates the license-free spectrum of 2.4 GHz (2.40-2.48 GHz), 5.2 GHz (5.15-5.35 GHz) and 5.8 GHz (5.725-5.825 GHz). WiMAX, based on the IEEE 802.16 standard, has been evaluated by companies for last mile connectivity, which can reach a theoretical up to 30 mile radius coverage. The WiMAX forum has published three licenses spectrum profiles, namely the 2.3 (2.3-2.4 GHz), 2.5 GHz (2.495-2.69 GHz) and 3.5 GHz (3.5-3.6 GHz) varying country to country. Many people expect WiMAX to emerge as another technology especially WiFi that may be adopted for handset devices and base station in the near future. The eleven standardized WiFi and WiMAX operating bands are listed in Table I.

Consequently, the research and manufacturing of both indoor and outdoor transmission equipment and devices fulfilling the requirements of these WiFi and WiMAX standards have increased since the idea took place in the technical and industrial community. An antenna serves as one of the critical component in any wireless communication system. As mentioned above, the design and development of a single antenna working in wideband or more frequency bands, called multiband antenna, is generally not an easy task. To answer these challenges, many antennas with wideband and/or multiband performances have been published in open literatures. The popular antenna for such applications is microstrip antenna (MSA) where several designs of multiband MSAs have been reported. Another important candidate, which may complete favorably with microstrip, is coplanar waveguide (CPW). Antennas using CPW-fed line also have many attractive features including lowradiation loss, less dispersion, easy integration for monolithic microwave circuits (MMICs) and a simple configuration with single metallic layer, since no backside processing is required for integration of devices. Therefore, the designs of CPW-fed antennas have recently become more and more attractive. One of the main issues with CPW-fed antennas is to provide an easy impedance matching to the CPW-fed line. In order to obtain multiband and broadband operations, several techniques have been reported in the literatures based on CPW-fed slot antennas (Chaimool et al., 2004, 2005, 2008; Sari-Kha et al., 2006; Jirasakulporn,

CPW-Fed Antennas for WiFi and WiMAX 21

as shown in Fig 1. Beside the microstrip line, the CPW is the most frequent use as planar transmission line in RF/microwave integrated circuits. It can be regarded as two coupled slot lines. Therefore, similar properties of a slot line may be expected. The CPW consists of three conductors with the exterior ones used as ground plates. These need not necessarily have same potential. As known from transmission line theory of a three-wire system, even and odd mode solutions exist as illustrated in Fig. 2. The desired even mode, also termed coplanar mode [Fig. 2 (a)] has ground electrodes at both sides of the centered strip, whereas the parasitic odd mode [Fig. 2 (b)], also termed slot line mode, has opposite electrode potentials. When the substrate is also metallized on its bottom side, an additional parasitic parallel plate mode with zero cutoff frequency can exist [Fig. 2(c)]. When a coplanar wave impinges on an asymmetric discontinuity such as a bend, parasitic slot line mode can be exited. To avoid these modes, bond wires or air bridges are connected to the ground places to force equal potential. Fig. 3 shows the electromagnetic field distribution of the even mode at low frequencies, which is TEM-like. At higher frequencies, the fundamental mode evolves itself approximately as a TE mode (H mode) with elliptical polarization of the magnetic field

(a) (b) (c)

(b) parasitic odd mode, and (c) parasitic parallel plate mode

Fig. 2. Schematic electrical field distribution in coplanar waveguide: (a) desired even mode,

Fig. 1. Coplanar waveguide structure (CPW)

in the slots.

2008), CPW-fed printed monopole (Chaimool et al., 2009; Moekham et al., 2011) and fractal techniques (Mahatthanajatuphat et al., 2009; Honghara et al., 2011).

In this chapter, a variety of advanced CPW-fed antenna designs suitable for WiFi and WiMAX operations is presented. Some promising CPW-fed slot antennas and CPW-fed monopole antenna to achieve bidirectional and/or omnidirectional with multiband operation are first shown. These antennas are suitable for practical portable devices. Then, in order to obtain the unidirectional radiation for base station antennas, CPW-fed slot antennas with modified shape reflectors have been proposed. By shaping the reflector, noticeable enhancements in both bandwidth and radiation pattern, which provides unidirectional radiation, can be achieved while maintaining the simple structure. This chapter is organized as follows. Section 2 provides the coplanar waveguide structure and characteristics. In section 3, the CPW-fed slot antennas with wideband operations are presented. The possibility of covering the standardized WiFi and WiMAX by using multiband CPW-fed slot antennas is explored in section 4. In order to obtain unidirectional radiation patterns, CPWfed slot antennas with modified reflectors and metasurface are designed and discussed in section 5. Finally, section 6 provides the concluding remarks.


Table 1. Designed operating bands and corresponding frequency ranges of WiFi and WiMAX

#### **2. Coplanar waveguide structure**

A coplanar waveguide (CPW) is a one type of strip transmission line defined as a planar transmission structure for transmitting microwave signals. It comprises of at least one flat conductive strip of small thickness, and conductive ground plates. A CPW structure consists of a median metallic strip of deposited on the surface of a dielectric substrate slab with two narrow slits ground electrodes running adjacent and parallel to the strip on the same surface

2008), CPW-fed printed monopole (Chaimool et al., 2009; Moekham et al., 2011) and fractal

In this chapter, a variety of advanced CPW-fed antenna designs suitable for WiFi and WiMAX operations is presented. Some promising CPW-fed slot antennas and CPW-fed monopole antenna to achieve bidirectional and/or omnidirectional with multiband operation are first shown. These antennas are suitable for practical portable devices. Then, in order to obtain the unidirectional radiation for base station antennas, CPW-fed slot antennas with modified shape reflectors have been proposed. By shaping the reflector, noticeable enhancements in both bandwidth and radiation pattern, which provides unidirectional radiation, can be achieved while maintaining the simple structure. This chapter is organized as follows. Section 2 provides the coplanar waveguide structure and characteristics. In section 3, the CPW-fed slot antennas with wideband operations are presented. The possibility of covering the standardized WiFi and WiMAX by using multiband CPW-fed slot antennas is explored in section 4. In order to obtain unidirectional radiation patterns, CPWfed slot antennas with modified reflectors and metasurface are designed and discussed in

System Designed Operating Bands Frequency Range (GHz)

2.4 GHz 2.4-2.485

2.3 GHz 2.3-2.4 2.5 GHz 2.5-2.69 3.3 GHz 3.3-3.4 3.5 GHz 3.4-3.6 3.7 GHz 3.6-3.8

3.7 GHz 3.6-3.8 5.8 GHz 5.725-5.850

5.2 GHz 5.15-5.35 5.5 GHz 5.47-5.725 5.8 GHz 5.725-5.875

techniques (Mahatthanajatuphat et al., 2009; Honghara et al., 2011).

section 5. Finally, section 6 provides the concluding remarks.

5 GHz

Table 1. Designed operating bands and corresponding frequency ranges of WiFi and

A coplanar waveguide (CPW) is a one type of strip transmission line defined as a planar transmission structure for transmitting microwave signals. It comprises of at least one flat conductive strip of small thickness, and conductive ground plates. A CPW structure consists of a median metallic strip of deposited on the surface of a dielectric substrate slab with two narrow slits ground electrodes running adjacent and parallel to the strip on the same surface

WiFi IEEE 802.11

Mobile WiMAX IEEE 802.16 2005

Fixed WiMAX IEEE 802.16 2004

**2. Coplanar waveguide structure** 

WiMAX

Fig. 1. Coplanar waveguide structure (CPW)

as shown in Fig 1. Beside the microstrip line, the CPW is the most frequent use as planar transmission line in RF/microwave integrated circuits. It can be regarded as two coupled slot lines. Therefore, similar properties of a slot line may be expected. The CPW consists of three conductors with the exterior ones used as ground plates. These need not necessarily have same potential. As known from transmission line theory of a three-wire system, even and odd mode solutions exist as illustrated in Fig. 2. The desired even mode, also termed coplanar mode [Fig. 2 (a)] has ground electrodes at both sides of the centered strip, whereas the parasitic odd mode [Fig. 2 (b)], also termed slot line mode, has opposite electrode potentials. When the substrate is also metallized on its bottom side, an additional parasitic parallel plate mode with zero cutoff frequency can exist [Fig. 2(c)]. When a coplanar wave impinges on an asymmetric discontinuity such as a bend, parasitic slot line mode can be exited. To avoid these modes, bond wires or air bridges are connected to the ground places to force equal potential. Fig. 3 shows the electromagnetic field distribution of the even mode at low frequencies, which is TEM-like. At higher frequencies, the fundamental mode evolves itself approximately as a TE mode (H mode) with elliptical polarization of the magnetic field in the slots.

Fig. 2. Schematic electrical field distribution in coplanar waveguide: (a) desired even mode, (b) parasitic odd mode, and (c) parasitic parallel plate mode

CPW-Fed Antennas for WiFi and WiMAX 23

 (a) (b) (c) (d) Fig. 5. CPW-fed slots with (a)-(b) inductive coupling and (c)–(d) capacitive coupling

presented from our previous work (Chaimool, et. al., 2004, 2005).

a widened tuning stub and (b) photograph of the prototype

**stub** 

**3.1 CPW-fed square slot antenna using loading metallic strips and a widened tuning** 

The geometry and prototype of the proposed CPW-fed slot antenna with loading metallic strips and widen tuning stub is shown in Fig. 6(a) and Fig. 6(b), respectively. The proposed antenna is fabricated on an inexpensive FR4 substrate with thickness (h) of 1.6 mm and relatively permittivity (r) of 4.4. The printed square radiating slot has a side length of *Lout* and a width of G. A 50- CPW has a signal strip of width *Wf,* and a gap of spacing *g*  between the signal strip and the coplanar ground plane. The widened tuning stub with a length of *L* and a width of *W* is connected to the end of the CPW feed line. Two loading metallic strips of the same dimensions (length of *L1* and width of 2 mm) are designed to protrude from the top comers into the slot center. The spacing between the tuning stub and edge of the ground plane is *S*. In this design, the dimensions are chosen to be G =72 mm, and *Lout* = 44 mm. Two parameters of the tuning stub including *L* and *W* and the length of loading metallic strip (*L1*) will affect the broadband operation. The parametric study was

 (a) (b) Fig. 6. (a) geometry of the proposed CPW-fed slot antenna using loading metallic strips and

Fig. 3. Transversal electromagnetic field of even coplanar mode at low frequency

#### **3. Wideband CPW-fed slot antennas**

To realize and cover WiFi and WiMAX operation bands, there are three ways to design antennas including (i) using broadband/wideband or ultrawideband techniques, (ii) using multiband techniques, and (iii) combining wideband and multiband techniques. For wideband operation, planar slot antennas are more promising because of their simple structure, easy to fabricate and wide impedance bandwidth characteristics. In general, the wideband CPW-fed slot antennas can be developed by tuning their impedance values. Several impedance tuning techniques are studied in literatures by varying the slot geometries and/or tuning stubs as shown in Fig. 4 and Fig. 5. Various slot geometries have been carried out such as wide rectangular slot, circular slot, elliptical slot, bow-tie slot, and hexagonal slot. Moreover, the impedance tuning can be done by using coupling mechanisms, namely inductive and capacitive couplings as shown Fig. 5. For capacitively coupled slots, several tuning stubs have been used such as circular, triangular, rectangular, and fractal shapes. In this section, we present the wideband slot antennas using CPW feed line. There are three antennas for wideband operations: CPW-fed square slot antenna using loading metallic strips and a widened tuning stub, CPW-fed equilateral hexagonal slot antennas, and CPW-fed slot antennas with fractal stubs.

Fig. 4. CPW-fed slots with various slot geometries and tuning stubs (a) wide rectangular slot, (b) circular slot, (c) triangular slot, (d) bow-tie slot, and (e) rectangular slot with fractal tuning stub

Fig. 3. Transversal electromagnetic field of even coplanar mode at low frequency

To realize and cover WiFi and WiMAX operation bands, there are three ways to design antennas including (i) using broadband/wideband or ultrawideband techniques, (ii) using multiband techniques, and (iii) combining wideband and multiband techniques. For wideband operation, planar slot antennas are more promising because of their simple structure, easy to fabricate and wide impedance bandwidth characteristics. In general, the wideband CPW-fed slot antennas can be developed by tuning their impedance values. Several impedance tuning techniques are studied in literatures by varying the slot geometries and/or tuning stubs as shown in Fig. 4 and Fig. 5. Various slot geometries have been carried out such as wide rectangular slot, circular slot, elliptical slot, bow-tie slot, and hexagonal slot. Moreover, the impedance tuning can be done by using coupling mechanisms, namely inductive and capacitive couplings as shown Fig. 5. For capacitively coupled slots, several tuning stubs have been used such as circular, triangular, rectangular, and fractal shapes. In this section, we present the wideband slot antennas using CPW feed line. There are three antennas for wideband operations: CPW-fed square slot antenna using loading metallic strips and a widened tuning stub, CPW-fed equilateral hexagonal slot

(a) (b) (c) (d) (e)

Fig. 4. CPW-fed slots with various slot geometries and tuning stubs (a) wide rectangular slot, (b) circular slot, (c) triangular slot, (d) bow-tie slot, and (e) rectangular slot with fractal

**3. Wideband CPW-fed slot antennas** 

antennas, and CPW-fed slot antennas with fractal stubs.

tuning stub

Fig. 5. CPW-fed slots with (a)-(b) inductive coupling and (c)–(d) capacitive coupling

#### **3.1 CPW-fed square slot antenna using loading metallic strips and a widened tuning stub**

The geometry and prototype of the proposed CPW-fed slot antenna with loading metallic strips and widen tuning stub is shown in Fig. 6(a) and Fig. 6(b), respectively. The proposed antenna is fabricated on an inexpensive FR4 substrate with thickness (h) of 1.6 mm and relatively permittivity (r) of 4.4. The printed square radiating slot has a side length of *Lout* and a width of G. A 50- CPW has a signal strip of width *Wf,* and a gap of spacing *g*  between the signal strip and the coplanar ground plane. The widened tuning stub with a length of *L* and a width of *W* is connected to the end of the CPW feed line. Two loading metallic strips of the same dimensions (length of *L1* and width of 2 mm) are designed to protrude from the top comers into the slot center. The spacing between the tuning stub and edge of the ground plane is *S*. In this design, the dimensions are chosen to be G =72 mm, and *Lout* = 44 mm. Two parameters of the tuning stub including *L* and *W* and the length of loading metallic strip (*L1*) will affect the broadband operation. The parametric study was presented from our previous work (Chaimool, et. al., 2004, 2005).

Fig. 6. (a) geometry of the proposed CPW-fed slot antenna using loading metallic strips and a widened tuning stub and (b) photograph of the prototype

CPW-Fed Antennas for WiFi and WiMAX 25

The far-field radiation patterns of the proposed antenna with the largest operating bandwidth using the design parameters of *L*1 =16 mm, *W* = 36 mm, *L* =22.5 mm, and S = 0.5 mm have been then measured. Fig. 8 shows the plots of the radiation patterns measured in y-z and x-z planes at the frequencies of 1660 and 2800 MHz. It has been found that we can obtain

This section introduces a new CPW-fed square slot antenna with loading metallic strips and a widened tuning stub for broadband operation. The simulation and experimental results of the proposed antenna show the impedance bandwidth, determined by 10-dB return loss, larger than 67% of the center frequency. The proposed antenna can be applied for WiFi (2.4

(a)

(b)

Fig. 9 shows the geometry and the prototype of the CPW-fed hexagonal slot antenna. It is designed and built on an FR4 substrate with thickness (h) of 1.6 mm and relatively permittivity (r) of 4.4. The ground plane is chosen to be an equilateral hexagonal structure with outer radius (*R*o) and inner radius (*Ri*). A 50- CPW feed line consists of a metal strip of width (*Wf* ) and a gap (g). This feed line is used to excite the proposed antenna. The tuning stub has a length of *Lf* and a width of *Wf*. For our design, the key dimensions of the proposed antenna are initially chosen to be *R*o = 55 mm, *Ri* = 33 mm, *Wf* = 6.37 mm, and g =

Fig. 8. Measured radiation patterns in the y-z and x-z planes for the proposed (a) f = 1660

acceptable broadside radiation patterns.

MHz and (b) f = 2800 MHz

**3.2 CPW-fed equilateral hexagonal slot antenna** 

GHz) and WiMAX (2.3 and 2.5 GHz bands) operations.

The present design is to make the first CPW-fed slot antenna to form a wider operating bandwidth. Firstly, a CPW-fed line is designed with the strip width *Wf* of 6.37 mm and a gap width g of 0.5 mm, corresponding to the characteristic impedance of 50-. The design structure has been obtained with the optimal tuning stub length of *L* =22.5 mm, tuning stub width *W* = 36 mm, and length of loading metallic strips *L*1 = 16 mm to perform the broadband operation. The proposed antenna has been constructed (Fig. 6(b)) and then tested using a calibrated vector network analyzer. Measured result of return losses compared with the simulation is shown in Fig. 7.

Fig. 7. Measured and simulated return losses for tuning stub width *W* = 36 mm, *L* = 22.5 mm, *Lout* = 44 mm, G=72 mm, *L*1=l6 mm, *Wf*=6.37 mm, and g = 0.5 mm, and (a) narrow band, (b) wideband views

The present design is to make the first CPW-fed slot antenna to form a wider operating bandwidth. Firstly, a CPW-fed line is designed with the strip width *Wf* of 6.37 mm and a gap width g of 0.5 mm, corresponding to the characteristic impedance of 50-. The design structure has been obtained with the optimal tuning stub length of *L* =22.5 mm, tuning stub width *W* = 36 mm, and length of loading metallic strips *L*1 = 16 mm to perform the broadband operation. The proposed antenna has been constructed (Fig. 6(b)) and then tested using a calibrated vector network analyzer. Measured result of return losses compared with

(a)

(b)

Fig. 7. Measured and simulated return losses for tuning stub width *W* = 36 mm, *L* = 22.5 mm, *Lout* = 44 mm, G=72 mm, *L*1=l6 mm, *Wf*=6.37 mm, and g = 0.5 mm, and (a) narrow band,

the simulation is shown in Fig. 7.

(b) wideband views

The far-field radiation patterns of the proposed antenna with the largest operating bandwidth using the design parameters of *L*1 =16 mm, *W* = 36 mm, *L* =22.5 mm, and S = 0.5 mm have been then measured. Fig. 8 shows the plots of the radiation patterns measured in y-z and x-z planes at the frequencies of 1660 and 2800 MHz. It has been found that we can obtain acceptable broadside radiation patterns.

This section introduces a new CPW-fed square slot antenna with loading metallic strips and a widened tuning stub for broadband operation. The simulation and experimental results of the proposed antenna show the impedance bandwidth, determined by 10-dB return loss, larger than 67% of the center frequency. The proposed antenna can be applied for WiFi (2.4 GHz) and WiMAX (2.3 and 2.5 GHz bands) operations.

Fig. 8. Measured radiation patterns in the y-z and x-z planes for the proposed (a) f = 1660 MHz and (b) f = 2800 MHz

#### **3.2 CPW-fed equilateral hexagonal slot antenna**

Fig. 9 shows the geometry and the prototype of the CPW-fed hexagonal slot antenna. It is designed and built on an FR4 substrate with thickness (h) of 1.6 mm and relatively permittivity (r) of 4.4. The ground plane is chosen to be an equilateral hexagonal structure with outer radius (*R*o) and inner radius (*Ri*). A 50- CPW feed line consists of a metal strip of width (*Wf* ) and a gap (g). This feed line is used to excite the proposed antenna. The tuning stub has a length of *Lf* and a width of *Wf*. For our design, the key dimensions of the proposed antenna are initially chosen to be *R*o = 55 mm, *Ri* = 33 mm, *Wf* = 6.37 mm, and g =

CPW-Fed Antennas for WiFi and WiMAX 27

proposed antenna has an operational frequency range from 1.657 to 2.956 GHz or

This section presents design and implementation of the CPW-fed equilateral hexagonal slot antenna. The transmission line and ground-plane have been designed to be on the same plane with the antenna slot to be applicable for wideband operation. It is found that the proposed antenna is accessible to bandwidth about 55.39%, a very large bandwidth comparing with conventional microstrip antennas, which mostly provide 1-5 % bandwidth. The proposed antenna can be used for many wireless systems such as WiFi , WiMAX,

In this section, the CPW-fed slot antenna with tuning stub of fractal geometry will be investigated. The Minkowski fractal structure will be modified to create the fractal stub of the proposed antenna. The proposed antennas have been designed and fabricated on an inexpensive FR4 substrate of thickness h = 0.8 mm and relative permittivity r = 4.2. The first antenna consists of a rectangular stub or zero iteration of fractal model (0 iteration), which has dimension of 10 mm × 25 mm. It is fed by 50 CPW-fed line with the strip width and distance gap of 7.2 mm and 0.48 mm, respectively. In the process of studying the fractal geometry on stub, it is begun by using a fractal model to repeat on a rectangular patch stub for creating the first and second iterations of fractal geometry on the stub, as shown in Fig. 11. Then, the fractal stub is connected by 50 CPW-fed line. On the second iteration fractal stub of the antenna, the fraction of size between the center element and four around elements is 1.35 because this value is suitable for completely fitting to connect between the center element and four around elements. As shown in Fig. 12(a), the dimensions of the second iteration antenna are following: WT= 48 mm, LT= 50 mm, WS1 = 39.84 mm, LS1 = 20.6 mm, WS2 = 15.84 mm, LS2 = 19.28 mm, WS3 = 7.42 mm, LS3 = 7.72 mm, WA = 25 mm, LB = 10

bandwidth about 55% of the center frequency measured at higher 10 dB return loss.

GSM1800, GSM1900, and IMT-2000.

mm, WTR= 7.2 mm, and h = 0.8 mm.

Fig. 11. The fractal model for stubs with different geometry iterations

**3.3 CPW-fed slot antennas with fractal stubs** 

0.5 mm, then we have adjusted three parameters including *R*o, *Ri*, and *Lf* to obtain a broadband operation.

Fig. 9. (a) geometry of the proposed CPW-fed equilateral hexagonal slot antenna and (b) the prototype of the proposed antenna (Sari-Kha et al., 2005)

Fig. 10. Simulated and measured return losses of the CPW-fed equilateral hexagonal slot antenna with *Ro* = 55 mm, *Ri* = 33 mm, and *Lf* = 42.625 mm

The optimal dimensions have been used for building up the proposed antenna. Measured return loss using a vector network analyzer is now shown in Fig.10. As we can see that the measured return loss agrees well with simulation expectation. It is also seen that the

0.5 mm, then we have adjusted three parameters including *R*o, *Ri*, and *Lf* to obtain a

(a) (b) Fig. 9. (a) geometry of the proposed CPW-fed equilateral hexagonal slot antenna and (b) the

Fig. 10. Simulated and measured return losses of the CPW-fed equilateral hexagonal slot

The optimal dimensions have been used for building up the proposed antenna. Measured return loss using a vector network analyzer is now shown in Fig.10. As we can see that the measured return loss agrees well with simulation expectation. It is also seen that the

prototype of the proposed antenna (Sari-Kha et al., 2005)

antenna with *Ro* = 55 mm, *Ri* = 33 mm, and *Lf* = 42.625 mm

broadband operation.

proposed antenna has an operational frequency range from 1.657 to 2.956 GHz or bandwidth about 55% of the center frequency measured at higher 10 dB return loss.

This section presents design and implementation of the CPW-fed equilateral hexagonal slot antenna. The transmission line and ground-plane have been designed to be on the same plane with the antenna slot to be applicable for wideband operation. It is found that the proposed antenna is accessible to bandwidth about 55.39%, a very large bandwidth comparing with conventional microstrip antennas, which mostly provide 1-5 % bandwidth. The proposed antenna can be used for many wireless systems such as WiFi , WiMAX, GSM1800, GSM1900, and IMT-2000.

#### **3.3 CPW-fed slot antennas with fractal stubs**

In this section, the CPW-fed slot antenna with tuning stub of fractal geometry will be investigated. The Minkowski fractal structure will be modified to create the fractal stub of the proposed antenna. The proposed antennas have been designed and fabricated on an inexpensive FR4 substrate of thickness h = 0.8 mm and relative permittivity r = 4.2. The first antenna consists of a rectangular stub or zero iteration of fractal model (0 iteration), which has dimension of 10 mm × 25 mm. It is fed by 50 CPW-fed line with the strip width and distance gap of 7.2 mm and 0.48 mm, respectively. In the process of studying the fractal geometry on stub, it is begun by using a fractal model to repeat on a rectangular patch stub for creating the first and second iterations of fractal geometry on the stub, as shown in Fig. 11. Then, the fractal stub is connected by 50 CPW-fed line. On the second iteration fractal stub of the antenna, the fraction of size between the center element and four around elements is 1.35 because this value is suitable for completely fitting to connect between the center element and four around elements. As shown in Fig. 12(a), the dimensions of the second iteration antenna are following: WT= 48 mm, LT= 50 mm, WS1 = 39.84 mm, LS1 = 20.6 mm, WS2 = 15.84 mm, LS2 = 19.28 mm, WS3 = 7.42 mm, LS3 = 7.72 mm, WA = 25 mm, LB = 10 mm, WTR= 7.2 mm, and h = 0.8 mm.

Fig. 11. The fractal model for stubs with different geometry iterations

CPW-Fed Antennas for WiFi and WiMAX 29

Iteration 0 4.3 4.5 1.6 - 7.1 1.7 – 7.1 123 121

Iteration 1 3.8 4.0 1.6 – 5.9 1.7 – 6.3 112 115

Iteration 2 2.7 2.8 1.6 – 3.8 1.7 – 4.0 78 82

frequencies decrease as increasing the iteration for fractal stub. Typically, the increasing iteration in the conventional fractal structure affects to the widely bandwidth. However, these results have inverted because the electrical length on the edge of stub, which the stub in the general CPW-fed slot antenna was used to control the higher frequency band, is increased and produced by the fractal geometry. In Table 3, simulation results show the antenna gains at operating frequency of 1.8 GHz, 2.1 GHz, 2.45 GHz, and 3.5 GHz above 3dBi. As the higher operating frequency, the average antenna gains are about 2 dBi. The overall dimension of CPW-fed fabricated slot antennas with fractal stub is 48× 50 × 0.8 mm3, as illustrated in Fig. 12(b). The simulated and measured results of the proposed antennas are compared as shown in Fig. 13. It can be clearly found that the simulated and measured results are similarity. However, the measured results of the return loss bandwidth slightly shift to higher frequency band. The error results are occurred due to the problem in fabrication because the fractal geometry stubs need the accuracy shapes. Moreover, the radiation patterns of 0, 1st and 2nd iteration stubs of the antennas are similar, which are the bidirectional radiation patterns at two frequencies, 2.45 and 3.5

Table 2. Comparison of characteristic results with different iterations of fractal stubs.

Operating Frequency Antenna Gain (dBi)

1.8 GHz Sim. 3.1 3.1 3.1

2.1 GHz Sim. 3.3 3.3 3.3

2.45 GHz Sim. 3.3 3.3 3.3

3.5 GHz Sim. 3.5 3.5 3.3

5.2 GHz Sim. 1.8 2.2 N/A

5.8 GHz Sim. 1.8 2.4 N/A

6.9 GHz Sim. 2.2 N/A N/A

Sim. Mea. Sim. Mea. Sim. Mea.

Return Loss Bandwidth (RL ≥10 dB)

Iteration 0 Iteration 1 Iteration 2

Mea. 2.1 2.5 2.7

Mea. 2.3 2.1 2.3

Mea. 2.9 2.8 2.6

Mea. 1.6 1.5 1.3

Mea. 1.1 1.7 N/A

Mea. 1.3 2.2 N/A

Mea. 2.1 N/A N/A

BW (GHz) BW (%)

Center Frequency (GHz)

Antenna type

GHz, as depicted in Fig. 14.

Table 3. Summarized results of the antenna gains

Fig. 12. (a) Geometry of the proposed CPW-fed slot antenna with the 2nd iteration fractal stub and (b) photograph of the fabricated antenna

In order to study the effects of fractal geometry on the stub of the slot antenna, IE3D program is used to simulate the characteristics and frequency responses of the antennas. The simulated return loss results of the 1st and 2nd iterations are shown in Fig. 13 and expanded in Table 2. The results show that all of return loss bandwidth tendencies and center

Fig. 13. Simulated and measured return losses of the proposed antenna with different iterations of fractal stubs

 (a) (b)

Fig. 12. (a) Geometry of the proposed CPW-fed slot antenna with the 2nd iteration fractal

Fig. 13. Simulated and measured return losses of the proposed antenna with different

In order to study the effects of fractal geometry on the stub of the slot antenna, IE3D program is used to simulate the characteristics and frequency responses of the antennas. The simulated return loss results of the 1st and 2nd iterations are shown in Fig. 13 and expanded in Table 2. The results show that all of return loss bandwidth tendencies and center

stub and (b) photograph of the fabricated antenna

iterations of fractal stubs


Table 2. Comparison of characteristic results with different iterations of fractal stubs.

frequencies decrease as increasing the iteration for fractal stub. Typically, the increasing iteration in the conventional fractal structure affects to the widely bandwidth. However, these results have inverted because the electrical length on the edge of stub, which the stub in the general CPW-fed slot antenna was used to control the higher frequency band, is increased and produced by the fractal geometry. In Table 3, simulation results show the antenna gains at operating frequency of 1.8 GHz, 2.1 GHz, 2.45 GHz, and 3.5 GHz above 3dBi. As the higher operating frequency, the average antenna gains are about 2 dBi. The overall dimension of CPW-fed fabricated slot antennas with fractal stub is 48× 50 × 0.8 mm3, as illustrated in Fig. 12(b). The simulated and measured results of the proposed antennas are compared as shown in Fig. 13. It can be clearly found that the simulated and measured results are similarity. However, the measured results of the return loss bandwidth slightly shift to higher frequency band. The error results are occurred due to the problem in fabrication because the fractal geometry stubs need the accuracy shapes. Moreover, the radiation patterns of 0, 1st and 2nd iteration stubs of the antennas are similar, which are the bidirectional radiation patterns at two frequencies, 2.45 and 3.5 GHz, as depicted in Fig. 14.


Table 3. Summarized results of the antenna gains

CPW-Fed Antennas for WiFi and WiMAX 31

possibility of covering some the standardized WiFi and WiMAX frequency bands while

In this section, we will show that CPW-fed slot antennas presented in the previous section (Section 3.1) can also be designed to demonstrate a dual-band behavior. The first dual-band antenna topology that, we introduce in Fig. 15(a); consists of the inner rectangular slot antenna with dimensions of w*in*×*Lin* and the outer square slot (*Lout* ×*Lout*). The outer square slot is used to control the first or lower operating band. On the other hand, the inner slot of width is used to control the second or upper operating band. The second antenna as shown in Fig. 15(b) combines a tuning stub with dimensions of *Ws* ×*L*3 placed in the inner slot at its bottom edge. The tuning stub is used to control coupling between a CPW feed line and the inner rectangular slot. In the third antenna as shown in Fig. 15(c), another pair of loading metallic strips is added at the bottom inner slot corners with dimensions of 1 mm×*L*2. Referring to Fig. 15(a), if adding a rectangular slot at tuning stub with w*in*= 21 mm and *Lin*= 11 mm to the wideband antenna (Fig. 6(a)), an additional resonant mode at about 5.2 GHz is obtained. This resonant mode excited is primarily owing to an inner rectangular slot. This way the antenna becomes a dual-band one in which the separation between the two resonant frequencies is a function of the resonant length of the second resonant frequency, the length and width of the inner slot (*Lin* and w*in*). To achieve the desired dual band operation of the rest antennas, we can adjust the parameters, (*W*, *L*, *L*1) and (w*in*, *Ws*, *L*2, *L*3, *Lin*), of the outer and inner slots, respectively, to control the lower and upper operating bands of the proposed antennas. The measured return losses of the proposed antennas are shown in Fig. 16. It can be observed that the multiband characteristics can be obtained. The impedance bandwidths of the lower band for all antennas are slightly different, and on the other hand, the upper band has an impedance bandwidth of 1680 MHz (4840–6520 MHz) for antenna in Fig. 15(b), which covers the WiFi band at 5.2 GHz and 5.8 GHz band for WiMAX. To sum up, the measured results and the corresponding settings of the parameters are listed

(a) (b) (c)

Fig. 15. Dual-band CPW-fed slot antennas with inner rectangular slot (a) without loading strip and a tuning stub, (b) with top corner loading strips and a bottom tuning stub, and (c)

with bottom corner loading strips and a top tuning stub

**4.1 Dual-band CPW-fed slot antennas using loading metallic strips and a widened** 

cling to the class of simply-structured and compact antennas.

**tuning stub** 

Fig. 14. Measured radiation patterns of the proposed CPW-fed slot antennas with 0, 1st and 2nd iteration fractal stubs (a) 2450 MHz and (b) 3500 MHz

This section studies CPW-fed slot antennas with fractal stubs. The return loss bandwidth of the antenna is affected by the fractal stub. It has been found that the antenna bandwidth decreases when the iteration of fractal stub increases, which it will be opposite to the conventional fractal structures. In this study, fractal models with the 0, 1st and 2nd iterations have been employed, resulting in the return loss bandwidths to be 121%, 115%, and 82%, respectively. Moreover, the radiation patterns of the presented antenna are still bidirections and the average gains of antenna are above 2 dBi for all of fractal stub iterations. Results indicate an impedance bandwidth covering the band for WiFi, WiMAX, and IMT-2000.

#### **4. Multiband CPW-fed slot antennas**

Design of antennas operating in multiband allows the wireless devices to be used with only a single antenna for multiple wireless applications, and thus permits to reduce the size of the space required for antenna on the wireless equipment. In this section, we explore the

(a)

(b) Fig. 14. Measured radiation patterns of the proposed CPW-fed slot antennas with 0, 1st and

This section studies CPW-fed slot antennas with fractal stubs. The return loss bandwidth of the antenna is affected by the fractal stub. It has been found that the antenna bandwidth decreases when the iteration of fractal stub increases, which it will be opposite to the conventional fractal structures. In this study, fractal models with the 0, 1st and 2nd iterations have been employed, resulting in the return loss bandwidths to be 121%, 115%, and 82%, respectively. Moreover, the radiation patterns of the presented antenna are still bidirections and the average gains of antenna are above 2 dBi for all of fractal stub iterations. Results indicate an impedance bandwidth covering the band for WiFi, WiMAX, and IMT-2000.

Design of antennas operating in multiband allows the wireless devices to be used with only a single antenna for multiple wireless applications, and thus permits to reduce the size of the space required for antenna on the wireless equipment. In this section, we explore the

2nd iteration fractal stubs (a) 2450 MHz and (b) 3500 MHz

**4. Multiband CPW-fed slot antennas** 

possibility of covering some the standardized WiFi and WiMAX frequency bands while cling to the class of simply-structured and compact antennas.

#### **4.1 Dual-band CPW-fed slot antennas using loading metallic strips and a widened tuning stub**

In this section, we will show that CPW-fed slot antennas presented in the previous section (Section 3.1) can also be designed to demonstrate a dual-band behavior. The first dual-band antenna topology that, we introduce in Fig. 15(a); consists of the inner rectangular slot antenna with dimensions of w*in*×*Lin* and the outer square slot (*Lout* ×*Lout*). The outer square slot is used to control the first or lower operating band. On the other hand, the inner slot of width is used to control the second or upper operating band. The second antenna as shown in Fig. 15(b) combines a tuning stub with dimensions of *Ws* ×*L*3 placed in the inner slot at its bottom edge. The tuning stub is used to control coupling between a CPW feed line and the inner rectangular slot. In the third antenna as shown in Fig. 15(c), another pair of loading metallic strips is added at the bottom inner slot corners with dimensions of 1 mm×*L*2. Referring to Fig. 15(a), if adding a rectangular slot at tuning stub with w*in*= 21 mm and *Lin*= 11 mm to the wideband antenna (Fig. 6(a)), an additional resonant mode at about 5.2 GHz is obtained. This resonant mode excited is primarily owing to an inner rectangular slot. This way the antenna becomes a dual-band one in which the separation between the two resonant frequencies is a function of the resonant length of the second resonant frequency, the length and width of the inner slot (*Lin* and w*in*). To achieve the desired dual band operation of the rest antennas, we can adjust the parameters, (*W*, *L*, *L*1) and (w*in*, *Ws*, *L*2, *L*3, *Lin*), of the outer and inner slots, respectively, to control the lower and upper operating bands of the proposed antennas. The measured return losses of the proposed antennas are shown in Fig. 16. It can be observed that the multiband characteristics can be obtained. The impedance bandwidths of the lower band for all antennas are slightly different, and on the other hand, the upper band has an impedance bandwidth of 1680 MHz (4840–6520 MHz) for antenna in Fig. 15(b), which covers the WiFi band at 5.2 GHz and 5.8 GHz band for WiMAX. To sum up, the measured results and the corresponding settings of the parameters are listed

Fig. 15. Dual-band CPW-fed slot antennas with inner rectangular slot (a) without loading strip and a tuning stub, (b) with top corner loading strips and a bottom tuning stub, and (c) with bottom corner loading strips and a top tuning stub

CPW-Fed Antennas for WiFi and WiMAX 33

(a)

(b) Fig. 17. Measured radiation patterns of the proposed antennas in case of optimized antennas

By inserting a slot and metallic strips at the widened stub in a single layer and fed by coplanar waveguide (CPW) transmission line, novel dual-band and broadband operations are presented. The proposed antennas are designed to have dual-band operation suitable for applications WiFi (2.4 and 5 GHz bands) and WiMAX (2.3, 2.5 and 5.8 bands) bands. The dual-band antennas are simple in design, and the two operating modes of the proposed antennas are associated with perimeter of slots and loading metallic strips, in which the lower operating band can be controlled by varying the perimeters of the outer square slot and the higher band depend on the inner slot of the widened stub. The experimental results of the proposed antennas show the impedance bandwidths of the two operating bands, determined

from 10-dB return loss, larger than 61% and 27% of the center frequencies, respectively.

in Table 4. (a) 1700 MHz, and (b) 5200 MHz

in Table 4. Radiation patterns of the proposed antennas were measured at two resonant frequencies. Fig. 17(a) and (b) show the y-*z* and *x*-*z* plane co- and cross-polarized patterns at 1700 and 5200 MHz, respectively. The radiation patterns are bidirectional on the broadside due to the outer slot mode at lower frequency and the radiation patterns are irregular because of the excitation of higher order mode, the traveling wave.

Fig. 16. Measured return losses of dual-band CPW-fed slot antennas


Table 4. Performance of the proposed dual-band CPW-fed slot antennas [Figs. 15(a), 15(b), and 15(c)] for different antenna parameter values of inner slot width (w*in*), length (*Lin*) and loading metallic strips in inner slot (*Ws*, *L*2, and *L*3) which *Lout* = 45 mm, *W* = 36 mm, *G*=72 mm, *L*1=16 mm, *L*= 22.5 mm, *h*=1.6 mm, *Wf* =6.37 mm, and g=0.5 mm

in Table 4. Radiation patterns of the proposed antennas were measured at two resonant frequencies. Fig. 17(a) and (b) show the y-*z* and *x*-*z* plane co- and cross-polarized patterns at 1700 and 5200 MHz, respectively. The radiation patterns are bidirectional on the broadside due to the outer slot mode at lower frequency and the radiation patterns are irregular

because of the excitation of higher order mode, the traveling wave.

Fig. 16. Measured return losses of dual-band CPW-fed slot antennas

7.5 6.0 11.0

> 20 20 20

> 20 20 20

mm, *L*1=16 mm, *L*= 22.5 mm, *h*=1.6 mm, *Wf* =6.37 mm, and g=0.5 mm

Fig. 15(a)

Fig. 15(b)

Fig. 15(c)

30 30 21

26 26 26

26 26 26 - - -

2 2 2

2 2 2

Dimension (mm) Bandwidth (S11 ≤ -10 dB)

61.0, 1600–3000 58.5, 1620–2960 58.2, 1630–2970

61.4, 1570–2960 49.4, 1600–2650 51.2, 1570–2650

58.7,1610–2950 57.8, 1610–2920 37.4, 1610–2350

7.5, 4880–5260 5.8, 5180–5490 16.1, 5040–5920

13.2, 5200–5935 10.0, 5305–5865 27.9, 5060–6705

9.3, 4900–5380 9.4, 4870–5350 10.0, 4840–5350

Antennas w*in WS Lin L2 L3* Lower BW(%,BW) Upper BW(%,BW)


6.0 8.0 10

7.0 9.0 11



9.5 9.5 9.5

Table 4. Performance of the proposed dual-band CPW-fed slot antennas [Figs. 15(a), 15(b), and 15(c)] for different antenna parameter values of inner slot width (w*in*), length (*Lin*) and loading metallic strips in inner slot (*Ws*, *L*2, and *L*3) which *Lout* = 45 mm, *W* = 36 mm, *G*=72

Fig. 17. Measured radiation patterns of the proposed antennas in case of optimized antennas in Table 4. (a) 1700 MHz, and (b) 5200 MHz

By inserting a slot and metallic strips at the widened stub in a single layer and fed by coplanar waveguide (CPW) transmission line, novel dual-band and broadband operations are presented. The proposed antennas are designed to have dual-band operation suitable for applications WiFi (2.4 and 5 GHz bands) and WiMAX (2.3, 2.5 and 5.8 bands) bands. The dual-band antennas are simple in design, and the two operating modes of the proposed antennas are associated with perimeter of slots and loading metallic strips, in which the lower operating band can be controlled by varying the perimeters of the outer square slot and the higher band depend on the inner slot of the widened stub. The experimental results of the proposed antennas show the impedance bandwidths of the two operating bands, determined from 10-dB return loss, larger than 61% and 27% of the center frequencies, respectively.

CPW-Fed Antennas for WiFi and WiMAX 35

Obviously, the achieved bandwidths not just cover the WiFi bands of 2.4 GHz (2.4-2.484 GHz) and 5.2 GHz (5.15-5.25 GHz), but also the licensed WiMAX bands of 2.5 GHz (2.5-2.69 GHz) and 3.5 GHz (3.4 -3.69 GHz). Fig. 20 shows the measured gains compared to the simulated result for all distinct bands. For the first two bands, gains are slightly decreased with frequency increases, whereas the gains in the upper band are fallen in with the simulation. The radiation characteristics have also been investigated and the measured patterns in two cuts (x-y plane, x-z plane) at 2.59, 3.52, and 5.98 GHz are plotted in Figs. 21(a), 21(b) and 21(c), respectively. As expected, the very good omni-directional patterns are obtained for all frequency bands in the x-y plane, whilst the close to bi-directional patterns

 (a) (b) Fig. 19. Photograph of the proposed CPW-fed mirrored-L monopole antenna (a) top layer

By coupling a stub-loaded open-loop resonator onto the back of a CPW-fed mirrored-L monopole, a novel triple-band planar antenna is achieved and presented in this section. The proposed antenna features a compact structure with reasonable gains. The measured bandwidths for the distinct triple-band are 2.27 to 2.87 GHz, 3.4 to 4.15 GHz and 5.11 to 6.7 GHz. Omni-directional radiation patterns for the three bands are observed. Simulations are confirmed by the experimental results, which ensure the proposed antenna is well suited for

in the x-z plane are observed.

and (b) bottom layer

the WiFi and WiMAX applications.

#### **4.2 CPW-fed mirrored-L monopole antenna with distinct triple bands**

Fig. 18 illustrates the geometry of the proposed triple-band antenna. A CPW-fed mirrored-L monopole is printed on one side (top layer) of an inexpensive FR4 dielectric substrate (dielectric constant *ε<sup>r</sup>* = 4.4, thickness *h* = 0.8 mm). An open-loop resonator loaded with an open stub is parasitically coupled on the back-side (bottom layer) of the mirrored-L monopole. The 50-Ω CPW feed line has a width of *wf* = 1.43 mm with gaps of g = 0.15 mm. Two symmetrical ground planes of size of 26 47 mm2 are used on the top layer. The openloop resonator has a length of about half-wavelength at 2.45 GHz but is loaded by an openstub of 4.6 mm. The unique resonator is responsible for the generation of resonant modes at 2.5 and 3.5 GHz, whereas the mirrored-L monopole joined with the feed-line is answerable for the wideband (5.11-6.7 GHz) generation. By properly tuning the relative positions (the coupling) between the L-shaped monopole and the open-loop resonator, and the spacing to the ground plane, the antenna exhibits three distinct bandwidths that fulfilling the required bandwidths from WiFi and WiMAX standards. Throughout the study, the IE3D simulator has been used for full-wave simulations in the design and optimization phases.

Fig. 18. Geometry of the proposed CPW-fed mirrored-L monopole antenna with dimensions in mm (a) top layer and (b) bottom layer

Based on the antenna parameters and the ground plane size depicted in Fig. 18, a prototype of this antenna was designed, fabricated and tested as shown in Fig. 19. Fig. 20 shows the measured return loss for the tri-band antenna. It is clearly seen that four resonant modes are excited at the frequencies of 2.59, 3.52, 5.56 and 6.37 GHz that results in three distinct bands. It is worthy of note that the latter two resonant modes are deliberately made in merge as a single wideband in order to cover all the unlicensed bands from 5.15 GHz to 5.85 GHz. The obtained 10-dB impedance bandwidths are 600 MHz (2.27-2.87 GHz), 750 MHz (3.4-4.15 GHz) and 1590 MHz (5.11-6.7 GHz), corresponding to the 23%, 20%, and 27%, respectively.

Fig. 18 illustrates the geometry of the proposed triple-band antenna. A CPW-fed mirrored-L monopole is printed on one side (top layer) of an inexpensive FR4 dielectric substrate (dielectric constant *ε<sup>r</sup>* = 4.4, thickness *h* = 0.8 mm). An open-loop resonator loaded with an open stub is parasitically coupled on the back-side (bottom layer) of the mirrored-L monopole. The 50-Ω CPW feed line has a width of *wf* = 1.43 mm with gaps of g = 0.15 mm. Two symmetrical ground planes of size of 26 47 mm2 are used on the top layer. The openloop resonator has a length of about half-wavelength at 2.45 GHz but is loaded by an openstub of 4.6 mm. The unique resonator is responsible for the generation of resonant modes at 2.5 and 3.5 GHz, whereas the mirrored-L monopole joined with the feed-line is answerable for the wideband (5.11-6.7 GHz) generation. By properly tuning the relative positions (the coupling) between the L-shaped monopole and the open-loop resonator, and the spacing to the ground plane, the antenna exhibits three distinct bandwidths that fulfilling the required bandwidths from WiFi and WiMAX standards. Throughout the study, the IE3D simulator

**4.2 CPW-fed mirrored-L monopole antenna with distinct triple bands** 

has been used for full-wave simulations in the design and optimization phases.

 (a) (b) Fig. 18. Geometry of the proposed CPW-fed mirrored-L monopole antenna with dimensions

Based on the antenna parameters and the ground plane size depicted in Fig. 18, a prototype of this antenna was designed, fabricated and tested as shown in Fig. 19. Fig. 20 shows the measured return loss for the tri-band antenna. It is clearly seen that four resonant modes are excited at the frequencies of 2.59, 3.52, 5.56 and 6.37 GHz that results in three distinct bands. It is worthy of note that the latter two resonant modes are deliberately made in merge as a single wideband in order to cover all the unlicensed bands from 5.15 GHz to 5.85 GHz. The obtained 10-dB impedance bandwidths are 600 MHz (2.27-2.87 GHz), 750 MHz (3.4-4.15 GHz) and 1590 MHz (5.11-6.7 GHz), corresponding to the 23%, 20%, and 27%, respectively.

in mm (a) top layer and (b) bottom layer

Obviously, the achieved bandwidths not just cover the WiFi bands of 2.4 GHz (2.4-2.484 GHz) and 5.2 GHz (5.15-5.25 GHz), but also the licensed WiMAX bands of 2.5 GHz (2.5-2.69 GHz) and 3.5 GHz (3.4 -3.69 GHz). Fig. 20 shows the measured gains compared to the simulated result for all distinct bands. For the first two bands, gains are slightly decreased with frequency increases, whereas the gains in the upper band are fallen in with the simulation. The radiation characteristics have also been investigated and the measured patterns in two cuts (x-y plane, x-z plane) at 2.59, 3.52, and 5.98 GHz are plotted in Figs. 21(a), 21(b) and 21(c), respectively. As expected, the very good omni-directional patterns are obtained for all frequency bands in the x-y plane, whilst the close to bi-directional patterns in the x-z plane are observed.

Fig. 19. Photograph of the proposed CPW-fed mirrored-L monopole antenna (a) top layer and (b) bottom layer

By coupling a stub-loaded open-loop resonator onto the back of a CPW-fed mirrored-L monopole, a novel triple-band planar antenna is achieved and presented in this section. The proposed antenna features a compact structure with reasonable gains. The measured bandwidths for the distinct triple-band are 2.27 to 2.87 GHz, 3.4 to 4.15 GHz and 5.11 to 6.7 GHz. Omni-directional radiation patterns for the three bands are observed. Simulations are confirmed by the experimental results, which ensure the proposed antenna is well suited for the WiFi and WiMAX applications.

CPW-Fed Antennas for WiFi and WiMAX 37

**4.3 Multiband antenna with modified fractal slot fed by CPW** 

 *<* 1. Normally, the appropriated value of iteration factor

11.11 mm, *W*3 = 16*.*05 mm, *W*4 = 3*.*7 mm, and *s*1 = *s*2 = *s*3 = 3.55 mm.

GHz, and (c) 5.98 GHz

(a) (b) (c)

varies with *Wp*. Usually, *Wp* is smaller than *Ws*/3 and the iteration factor is

Fig. 22. Measured far-field radiation patterns in *x-y* plane and *x-z* plane (a) 2.59 GHz, (b) 3.52

In this section, a fractal slot antenna fed by CPW was created by applying the Minkowski fractal concept to generate the initial generator model at both sides of inner patch of the antenna, as shown in Fig. 23. The altitude of initial generator model as shown in Fig. 24

fractal slot antenna. The configuration of the proposed antenna, as illustrated in Fig. 23, is the modified fractal slot antenna fed by CPW. The antenna composes of the modified inner metallic patch, which is fed by a 50-CPW line with a strip width *Wf* and gap *g*1, and an outer metallic patch. In the section, the antenna is fabricated on an economical FR4 dielectric substrate with a thickness of 1.6 mm (*h*), relative permittivity of 4.1 and loss tangent of 0.019. The entire dimensions of the antenna are 53.40mm *×* 75.20 mm. The 50- SMA connector is used to feed the antenna at the CPW line. The important parameters, which affect the resonant frequencies of 1.74 GHz, 3.85 GHz, and 5.05 GHz, compose of *Su*, *S*, and *SL*. The fixed parameters of the proposed antenna are following: *h* = 1.6 mm, *WG*1 = 53*.*37 mm, *WG*2 = 38.54 mm, *LG*1 = 75.20 mm, *LG*2 = 34.07 mm, *LG*3 = 39.75 mm, *Ws* = 32.57 mm, *g*1 = 0.5 mm, *g*<sup>2</sup> = 2.3 mm, *Wt* = 0.94 mm, *Lt* = 21.88 mm, *Wf* = 3.5 mm, *Lf* = 14.50 mm, *W*1 = 25.92 mm, *W*2 =

= 0.66 was used to produce the

= *3Wp/Ws*; 0 *<* 

Fig. 20. Measured return losses versus frequency

Fig. 21. Simulated and measured realized gains

Fig. 20. Measured return losses versus frequency

Fig. 21. Simulated and measured realized gains

Fig. 22. Measured far-field radiation patterns in *x-y* plane and *x-z* plane (a) 2.59 GHz, (b) 3.52 GHz, and (c) 5.98 GHz

#### **4.3 Multiband antenna with modified fractal slot fed by CPW**

In this section, a fractal slot antenna fed by CPW was created by applying the Minkowski fractal concept to generate the initial generator model at both sides of inner patch of the antenna, as shown in Fig. 23. The altitude of initial generator model as shown in Fig. 24 varies with *Wp*. Usually, *Wp* is smaller than *Ws*/3 and the iteration factor is = *3Wp/Ws*; 0 *< <* 1. Normally, the appropriated value of iteration factor = 0.66 was used to produce the fractal slot antenna. The configuration of the proposed antenna, as illustrated in Fig. 23, is the modified fractal slot antenna fed by CPW. The antenna composes of the modified inner metallic patch, which is fed by a 50-CPW line with a strip width *Wf* and gap *g*1, and an outer metallic patch. In the section, the antenna is fabricated on an economical FR4 dielectric substrate with a thickness of 1.6 mm (*h*), relative permittivity of 4.1 and loss tangent of 0.019. The entire dimensions of the antenna are 53.40mm *×* 75.20 mm. The 50- SMA connector is used to feed the antenna at the CPW line. The important parameters, which affect the resonant frequencies of 1.74 GHz, 3.85 GHz, and 5.05 GHz, compose of *Su*, *S*, and *SL*. The fixed parameters of the proposed antenna are following: *h* = 1.6 mm, *WG*1 = 53*.*37 mm, *WG*2 = 38.54 mm, *LG*1 = 75.20 mm, *LG*2 = 34.07 mm, *LG*3 = 39.75 mm, *Ws* = 32.57 mm, *g*1 = 0.5 mm, *g*<sup>2</sup> = 2.3 mm, *Wt* = 0.94 mm, *Lt* = 21.88 mm, *Wf* = 3.5 mm, *Lf* = 14.50 mm, *W*1 = 25.92 mm, *W*2 = 11.11 mm, *W*3 = 16*.*05 mm, *W*4 = 3*.*7 mm, and *s*1 = *s*2 = *s*3 = 3.55 mm.

CPW-Fed Antennas for WiFi and WiMAX 39

The suitable parameters, as following, *h* = 1.6 mm, *WG*1 = 53.37 mm, *WG*2 = 38*.*54 mm, *LG*1 = 75*.*20 mm, *LG*2 = 34*.*07 mm, *LG*3 = 39*.*75 mm, *Ws* = 32*.*57 mm, *g*1 = 0*.*5 mm, *g*2 = 2*.*3 mm, *Wt* = 0*.*94 mm, *Lt* = 21*.*88 mm, *Wf* = 3*.*5 mm, *Lf* = 14*.*50 mm, *W*1 = 25*.*92 mm, *W*2 = 11*.*11 mm, *W*3 = 16*.*05 mm, *W*4 = 3*.*7 mm, and *s*1 = *s*2 = *s*3 = 3*.*55 mm, *Su* = 16*.*050 mm, *S* = 4*.*751 mm, and *SL* = 16*.*050 mm, are chosen to implement the prototype antenna by etching into chemicals. The prototype of the proposed antenna is shown in Fig. 23(b). The simulated and measured return losses of the antenna are illustrated in Fig. 25. It is clearly observed that the measured return loss of the antenna slightly shifts to the right because of the inaccuracy of the manufacturing process by etching into chemicals. However, the measured result of proposed antenna still covers the operating bands of 1.71-1.88 GHz and 3.2-5.5 GHz for the

This section presents a multiband slot antenna with modifying fractal geometry fed by CPW transmission line. The presented antenna has been designed by modifying an inner fractal patch of the antenna to operate at multiple resonant frequencies, which effectively supports the digital communication system (DCS1800 1.71-1.88 GHz), WiMAX (3.30-3.80 GHz), and WiFi (5.15-5.35 GHz). Manifestly, it has been found that the radiation patterns of the presented antenna are still similarly to the bidirectional radiation pattern at all operating frequencies.

applications of DCS 1800, WiMAX (3.3 and 3.5 bands), and WiFi (5.5 GHz band).

Fig. 25. Simulated and measured return losses for the proposed antenna

Fig. 24. The initial generator model for the proposed antenna

(b)

Fig. 23. (a) Configurations of the proposed fractal slot antenna and (b) photograph of the prototype

(a)

(b)

Fig. 23. (a) Configurations of the proposed fractal slot antenna and (b) photograph of the

prototype

Fig. 24. The initial generator model for the proposed antenna

The suitable parameters, as following, *h* = 1.6 mm, *WG*1 = 53.37 mm, *WG*2 = 38*.*54 mm, *LG*1 = 75*.*20 mm, *LG*2 = 34*.*07 mm, *LG*3 = 39*.*75 mm, *Ws* = 32*.*57 mm, *g*1 = 0*.*5 mm, *g*2 = 2*.*3 mm, *Wt* = 0*.*94 mm, *Lt* = 21*.*88 mm, *Wf* = 3*.*5 mm, *Lf* = 14*.*50 mm, *W*1 = 25*.*92 mm, *W*2 = 11*.*11 mm, *W*3 = 16*.*05 mm, *W*4 = 3*.*7 mm, and *s*1 = *s*2 = *s*3 = 3*.*55 mm, *Su* = 16*.*050 mm, *S* = 4*.*751 mm, and *SL* = 16*.*050 mm, are chosen to implement the prototype antenna by etching into chemicals. The prototype of the proposed antenna is shown in Fig. 23(b). The simulated and measured return losses of the antenna are illustrated in Fig. 25. It is clearly observed that the measured return loss of the antenna slightly shifts to the right because of the inaccuracy of the manufacturing process by etching into chemicals. However, the measured result of proposed antenna still covers the operating bands of 1.71-1.88 GHz and 3.2-5.5 GHz for the applications of DCS 1800, WiMAX (3.3 and 3.5 bands), and WiFi (5.5 GHz band).

This section presents a multiband slot antenna with modifying fractal geometry fed by CPW transmission line. The presented antenna has been designed by modifying an inner fractal patch of the antenna to operate at multiple resonant frequencies, which effectively supports the digital communication system (DCS1800 1.71-1.88 GHz), WiMAX (3.30-3.80 GHz), and WiFi (5.15-5.35 GHz). Manifestly, it has been found that the radiation patterns of the presented antenna are still similarly to the bidirectional radiation pattern at all operating frequencies.

Fig. 25. Simulated and measured return losses for the proposed antenna

CPW-Fed Antennas for WiFi and WiMAX 41

plate is a useful modification of the corner reflector. To reduce overall dimensions of a large corner reflector, the vertex can be cut off and replaced with the horizontal flat reflector (Wc1×Wc3). The geometry of the proposed wideband CPW-fed slot antenna using -shaped reflector with the horizontal plate is shown in Fig. 27(c). The -shaped reflector, having a horizontal flat section dimension of Wc1×Wc3, is bent with a bent angle of . The width of the bent section of the -shaped reflector is Wc2. The distance between the antenna and the flat section is hc. For the last reflector, we modified the conductor reflector shape. Instead of the -shaped reflector, we took the conductor reflector to have the form of an inverted shaped reflector. The geometry of the inverted -shaped reflector with the horizontal plate is shown in Fig. 27(d). The inverted -shaped reflector, having a horizontal flat section dimension of Wd1×Wd3, is bent with a bent angle of . The width of the bent section of the inverted -shaped reflector is Wd2. The distance between the antenna and the flat section is hd. Several parameters have been reported in (Akkaraekthalin et al., 2007). In this section, three typical cases are investigated: (i) the -shaped reflector with hc = 30 mm, =150°, Wc1= 200 mm, Wc2 = 44 mm, beamwidth in H-plane around 72°, as called **72 DegAnt**; (ii) the shaped reflector with hc = 30 mm, =150°, Wc1 = 72 mm, Wc2 = 44 mm, beamwidth in Hplane around 90°, as called **90 DegAnt**; and (iii) the inverted -shaped reflector with hd = 50 mm, = 120°, Wd1 = 72 mm, Wd2 = 44 mm, beamwidth in H-plane around 120°, as called **120 DegAnt.** The prototypes of the proposed antennas were constructed as shown in Fig. 28. Fig. 29 shows the measured return losses of the proposed antenna. The 10-dB bandwidth is about 69% (1.5 to 3.1 GHz) of 72DegAnt. A very wide impedance bandwidth of 73% (1.5 - 3.25 GHz) for the antenna of 90DegAnt was achieved. The last, impedance bandwidth is 49% (1.88 to 3.12 GHz) when the antenna is 120DegAnt as shown in Fig. 29. However, from the obtained results of the three antennas, it is clearly seen that the broadband bandwidth for PCS/DCS/IMT-2000 WiFi and WiMAX bands is obtained. The radiation characteristics are also investigated. Fig. 30 presents the measured far-field radiation patterns of the proposed antennas at 1800 MHz, 2400 MHz, and 2800 MHz. As expected, the reflectors allow the antennas to radiate unidirectionally, the antennas keep the similar radiation patterns at several separated selected frequencies. The radiation patterns are stable across the matched frequency band. The main beams of normalized H-plane patterns at 1.8, 2.4, and 2.8 GHz are also measured for three different reflector shapes as shown in Fig. 31. Finally, the measured antenna gains in the broadside direction is presented in Fig. 32. For the 72DegAnt, the measured antenna gain is about 7.0 dBi over the entire viable frequency band.

Fig. 27. CPW-FSLW (a) radiating element above, (b) at reector, (c) -shaped reector with a horizontal plate, and (d) inverted -shaped reector with a horizontal plate

#### **5. Unidirectional CPW-fed slot antennas**

From the previous sections, most of the proposed antennas have bidirectional radiation patterns, with the back radiation being undesired directions but also increases the sensitivity of the antenna to its surrounding environment and prohibits the placement of such slot antennas on the platforms. A CPW-fed slot antenna naturally radiates bidirectionally, this characteristic is necessary for some applications, such as antennas for roads. However, this inherent bidirectional radiation is undesired in some wireless communication applications such as in base station antenna. There are several methods in order to reduce backside radiation and increase the gain. Two common approaches are to add an additional metal reflector and an enclosed cavity underneath the slot to redirect radiated energy from an undesired direction. In this section, promising wideband CPW-fed slot antennas with unidirectional radiation pattern developed for WiFi and WiMAX applications are presented. We propose two techniques for redirect the back radiation forward including (i) using modified the reflectors placed underneath the slot antennas (Fig. 26(a)) and (ii) the new technique by using the metasurface as a superstrate as shown in Fig 26(b).

Fig. 26. Arrangement of unidirectional CPW-fed slot antennas (a) conventional structure using conductor-back reflector and (b) the proposed structure using metasurface superstrate

#### **5.1 Wideband unidirectional CPW-fed slot antenna using loading metallic strips and a widened tuning stub**

The geometry of a CPW-fed slot antennas using loading metallic strips and a widened tuning stub is depicted in Fig. 27(a). Three different geometries of the proposed conducting reector behind CPW-fed slot antennas using loading metallic strips and a widened tuning stub are shown in Figs. 27(b), (c), and (d). It comprises of a single FR4 layer suspended over a metallic reector, which allows to use a single substrate and to minimize wiring and soldering. The antenna is designed on a FR4 substrate 1.6 mm thick, with relative dielectric constant (r) 4.4. This structure without a reector radiates a bidirectional pattern and maximum gain is about 4.5 dBi. The rst antenna, Fig. 27(b), is the antenna located above a at reector, with a reector size 100×100 mm2. The -shaped reflector with the horizontal

From the previous sections, most of the proposed antennas have bidirectional radiation patterns, with the back radiation being undesired directions but also increases the sensitivity of the antenna to its surrounding environment and prohibits the placement of such slot antennas on the platforms. A CPW-fed slot antenna naturally radiates bidirectionally, this characteristic is necessary for some applications, such as antennas for roads. However, this inherent bidirectional radiation is undesired in some wireless communication applications such as in base station antenna. There are several methods in order to reduce backside radiation and increase the gain. Two common approaches are to add an additional metal reflector and an enclosed cavity underneath the slot to redirect radiated energy from an undesired direction. In this section, promising wideband CPW-fed slot antennas with unidirectional radiation pattern developed for WiFi and WiMAX applications are presented. We propose two techniques for redirect the back radiation forward including (i) using modified the reflectors placed underneath the slot antennas (Fig. 26(a)) and (ii) the new

technique by using the metasurface as a superstrate as shown in Fig 26(b).

Fig. 26. Arrangement of unidirectional CPW-fed slot antennas (a) conventional structure using conductor-back reflector and (b) the proposed structure using metasurface superstrate

**5.1 Wideband unidirectional CPW-fed slot antenna using loading metallic strips and a** 

The geometry of a CPW-fed slot antennas using loading metallic strips and a widened tuning stub is depicted in Fig. 27(a). Three different geometries of the proposed conducting reector behind CPW-fed slot antennas using loading metallic strips and a widened tuning stub are shown in Figs. 27(b), (c), and (d). It comprises of a single FR4 layer suspended over a metallic reector, which allows to use a single substrate and to minimize wiring and soldering. The antenna is designed on a FR4 substrate 1.6 mm thick, with relative dielectric constant (r) 4.4. This structure without a reector radiates a bidirectional pattern and maximum gain is about 4.5 dBi. The rst antenna, Fig. 27(b), is the antenna located above a at reector, with a reector size 100×100 mm2. The -shaped reflector with the horizontal

**5. Unidirectional CPW-fed slot antennas** 

**widened tuning stub** 

plate is a useful modification of the corner reflector. To reduce overall dimensions of a large corner reflector, the vertex can be cut off and replaced with the horizontal flat reflector (Wc1×Wc3). The geometry of the proposed wideband CPW-fed slot antenna using -shaped reflector with the horizontal plate is shown in Fig. 27(c). The -shaped reflector, having a horizontal flat section dimension of Wc1×Wc3, is bent with a bent angle of . The width of the bent section of the -shaped reflector is Wc2. The distance between the antenna and the flat section is hc. For the last reflector, we modified the conductor reflector shape. Instead of the -shaped reflector, we took the conductor reflector to have the form of an inverted shaped reflector. The geometry of the inverted -shaped reflector with the horizontal plate is shown in Fig. 27(d). The inverted -shaped reflector, having a horizontal flat section dimension of Wd1×Wd3, is bent with a bent angle of . The width of the bent section of the inverted -shaped reflector is Wd2. The distance between the antenna and the flat section is hd. Several parameters have been reported in (Akkaraekthalin et al., 2007). In this section, three typical cases are investigated: (i) the -shaped reflector with hc = 30 mm, =150°, Wc1= 200 mm, Wc2 = 44 mm, beamwidth in H-plane around 72°, as called **72 DegAnt**; (ii) the shaped reflector with hc = 30 mm, =150°, Wc1 = 72 mm, Wc2 = 44 mm, beamwidth in Hplane around 90°, as called **90 DegAnt**; and (iii) the inverted -shaped reflector with hd = 50 mm, = 120°, Wd1 = 72 mm, Wd2 = 44 mm, beamwidth in H-plane around 120°, as called **120 DegAnt.** The prototypes of the proposed antennas were constructed as shown in Fig. 28. Fig. 29 shows the measured return losses of the proposed antenna. The 10-dB bandwidth is about 69% (1.5 to 3.1 GHz) of 72DegAnt. A very wide impedance bandwidth of 73% (1.5 - 3.25 GHz) for the antenna of 90DegAnt was achieved. The last, impedance bandwidth is 49% (1.88 to 3.12 GHz) when the antenna is 120DegAnt as shown in Fig. 29. However, from the obtained results of the three antennas, it is clearly seen that the broadband bandwidth for PCS/DCS/IMT-2000 WiFi and WiMAX bands is obtained. The radiation characteristics are also investigated. Fig. 30 presents the measured far-field radiation patterns of the proposed antennas at 1800 MHz, 2400 MHz, and 2800 MHz. As expected, the reflectors allow the antennas to radiate unidirectionally, the antennas keep the similar radiation patterns at several separated selected frequencies. The radiation patterns are stable across the matched frequency band. The main beams of normalized H-plane patterns at 1.8, 2.4, and 2.8 GHz are also measured for three different reflector shapes as shown in Fig. 31. Finally, the measured antenna gains in the broadside direction is presented in Fig. 32. For the 72DegAnt, the measured antenna gain is about 7.0 dBi over the entire viable frequency band.

Fig. 27. CPW-FSLW (a) radiating element above, (b) at reector, (c) -shaped reector with a horizontal plate, and (d) inverted -shaped reector with a horizontal plate

CPW-Fed Antennas for WiFi and WiMAX 43

(a) (b) (c)

(a) (b) (c)

Fig. 31. Measured radiation patterns in H-plane for three different reflectors at (a) 1800

(90DegAnt), and (c) 120° (120DegAnt) (Chaimool et al., 2011)

MHz, (b) 2400 MHz, and (c) 2800 MHz (Chaimool et al., 2011)

Fig. 32. Measured gains of the fabricated antennas

Fig. 30. Measured radiation pattern of three different reflectors, (a) 72° (72DegAnt), (b) 90°

As shown, the gain variations are smooth. The average gains of the 90DegAnt and 120DegAnt over this bandwidth are 6 dBi and 5 dBi, respectively. This is due to impedance mismatch and pattern degradation, as the back radiation level increases rapidly at these frequencies.

Fig. 28. Photograph of the fabricated antennas (Akkaraekthalin et al., 2007)

Fig. 29. Measured return losses of three different reflectors :72° (72DegAnt), 90° (90DegAnt), and 120° (120DegAnt)

As shown, the gain variations are smooth. The average gains of the 90DegAnt and 120DegAnt over this bandwidth are 6 dBi and 5 dBi, respectively. This is due to impedance mismatch and

pattern degradation, as the back radiation level increases rapidly at these frequencies.

Fig. 28. Photograph of the fabricated antennas (Akkaraekthalin et al., 2007)

Fig. 29. Measured return losses of three different reflectors :72° (72DegAnt), 90° (90DegAnt),

and 120° (120DegAnt)

Fig. 30. Measured radiation pattern of three different reflectors, (a) 72° (72DegAnt), (b) 90° (90DegAnt), and (c) 120° (120DegAnt) (Chaimool et al., 2011)

Fig. 31. Measured radiation patterns in H-plane for three different reflectors at (a) 1800 MHz, (b) 2400 MHz, and (c) 2800 MHz (Chaimool et al., 2011)

Fig. 32. Measured gains of the fabricated antennas

CPW-Fed Antennas for WiFi and WiMAX 45

(a)

(b) Fig. 34. (a) Photograph of the prototype antenna and (b) simulated and measured S11 of the

An improvement in the gain of 6.5 dB has been obtained. It is obtained that the realized

(a) (b) (c) Fig. 35. Measured radiation patterns for the CPW-fed slot antenna with the metasurface in

gains of the present metasurface are all improved within the operating bandwidth.

CPW-fed slot antenna with the metasurface (Rakluea et al. 2011)

*E-*plane. (a) 2400 MHz, (b) 2450 MHz and (c) 2500 MHz

#### **5.2 Unidirectional CPW-fed slot antenna using metasurface**

Fig. 33 shows the configurations of the proposed antenna. It consists of a CPW-fed slot antenna beneath a metasurface with the air-gap separation *ha.* The radiator is center-fed inductively coupled slot, where the slot has a length (*L-Wf* ) and width W. A 50- CPW transmission line, having a signal strip of width *Wf* and a gap of distance g, is used to excite the slot. The slot length determines the resonant length, while the slot width can be adjusted to achieve a wider bandwidth. The antenna is printed on 1.6 mm thick (h1) FR4 material with a dielectric constant (r1) of 4.2. For the metasurface as shown in Fig. 33(b), it comprises of an array 4×4 square loop resonators (SLRs). It is printed on an inexpensive FR4 substrate with dielectric constant r2= 4.2 and thickness (h2) 0.8 mm. The physical parameters of the SLR are given as follows: P = 20 mm, a = 19 mm and b= 18 mm. To validate the proposed concept, a prototype of the CPW-fed slot antenna with metasurface was designed, fabricated and measured as shown in Fig. 34 (a). The metasurface is supported by four plastic posts above the CPW-fed slot antenna with *ha* = 6.0 mm, having dimensions of 108 mm´108 mm (0.86<sup>0</sup> ´0.860). Simulations were conducted by using IE3D simulator, a full-wave moment-ofmethod (MoM) solver, and its characteristics were measured by a vector network analyzer. The S11 obtained from simulation and measurement of the CPW-fed slot antenna with metasurface with a very good agreement is shown in Fig. 34 (b). The measured impedance bandwidth (S11 ≤ -10 dB) is from 2350 to 2600 MHz (250 MHz or 10%). The obtained bandwidth covers the required bandwidth of the WiFi and WiMAX systems (2300-2500 MHz). Some errors in the resonant frequency occurred due to tolerance in FR4 substrate and poor manufacturing in the laboratory. Corresponding radiation patterns and realized gains of the proposed antenna were measured in the anechoic antenna chamber located at the Rajamangala University of Technology Thanyaburi (RMUTT), Thailand. The measured radiation patterns at 2400, 2450 and 2500 MHz with both co- and cross-polarization in *E*- and *H*- planes are given in Fig. 35 and 36, respectively. Very good broadside patterns are observed and the cross-polarization in the principal planes is seen to be than -20 dB for all of the operating frequency. The front-to-back ratios FBRs were also measured. From measured results, the FBRs are more than 15 and 10 dB for *E*- and *H*- planes, respectively. Moreover, the realized gains of the CPW-fed slot antenna with and without the metasurface were measured and compared as shown in Fig. 37. The gain for absence metasurface is about 1.5 dBi, whereas the presence metasurface can increase to 8.0 dBi at the center frequency.

Fig. 33. Configuration of the CPW-fed slot antenna with metasurface (a) the CPW-fed slot antenna, (b) metasurface and (c) the cross sectional view

Fig. 33 shows the configurations of the proposed antenna. It consists of a CPW-fed slot antenna beneath a metasurface with the air-gap separation *ha.* The radiator is center-fed inductively coupled slot, where the slot has a length (*L-Wf* ) and width W. A 50- CPW transmission line, having a signal strip of width *Wf* and a gap of distance g, is used to excite the slot. The slot length determines the resonant length, while the slot width can be adjusted to achieve a wider bandwidth. The antenna is printed on 1.6 mm thick (h1) FR4 material with a dielectric constant (r1) of 4.2. For the metasurface as shown in Fig. 33(b), it comprises of an array 4×4 square loop resonators (SLRs). It is printed on an inexpensive FR4 substrate with dielectric constant r2= 4.2 and thickness (h2) 0.8 mm. The physical parameters of the SLR are given as follows: P = 20 mm, a = 19 mm and b= 18 mm. To validate the proposed concept, a prototype of the CPW-fed slot antenna with metasurface was designed, fabricated and measured as shown in Fig. 34 (a). The metasurface is supported by four plastic posts above the CPW-fed slot antenna with *ha* = 6.0 mm, having dimensions of 108 mm´108 mm (0.86<sup>0</sup> ´0.860). Simulations were conducted by using IE3D simulator, a full-wave moment-ofmethod (MoM) solver, and its characteristics were measured by a vector network analyzer. The S11 obtained from simulation and measurement of the CPW-fed slot antenna with metasurface with a very good agreement is shown in Fig. 34 (b). The measured impedance bandwidth (S11 ≤ -10 dB) is from 2350 to 2600 MHz (250 MHz or 10%). The obtained bandwidth covers the required bandwidth of the WiFi and WiMAX systems (2300-2500 MHz). Some errors in the resonant frequency occurred due to tolerance in FR4 substrate and poor manufacturing in the laboratory. Corresponding radiation patterns and realized gains of the proposed antenna were measured in the anechoic antenna chamber located at the Rajamangala University of Technology Thanyaburi (RMUTT), Thailand. The measured radiation patterns at 2400, 2450 and 2500 MHz with both co- and cross-polarization in *E*- and *H*- planes are given in Fig. 35 and 36, respectively. Very good broadside patterns are observed and the cross-polarization in the principal planes is seen to be than -20 dB for all of the operating frequency. The front-to-back ratios FBRs were also measured. From measured results, the FBRs are more than 15 and 10 dB for *E*- and *H*- planes, respectively. Moreover, the realized gains of the CPW-fed slot antenna with and without the metasurface were measured and compared as shown in Fig. 37. The gain for absence metasurface is about 1.5 dBi, whereas

**5.2 Unidirectional CPW-fed slot antenna using metasurface** 

the presence metasurface can increase to 8.0 dBi at the center frequency.

(a) (b) (c)

antenna, (b) metasurface and (c) the cross sectional view

Fig. 33. Configuration of the CPW-fed slot antenna with metasurface (a) the CPW-fed slot

Fig. 34. (a) Photograph of the prototype antenna and (b) simulated and measured S11 of the CPW-fed slot antenna with the metasurface (Rakluea et al. 2011)

An improvement in the gain of 6.5 dB has been obtained. It is obtained that the realized gains of the present metasurface are all improved within the operating bandwidth.

Fig. 35. Measured radiation patterns for the CPW-fed slot antenna with the metasurface in *E-*plane. (a) 2400 MHz, (b) 2450 MHz and (c) 2500 MHz

CPW-Fed Antennas for WiFi and WiMAX 47

of antennas are fabricated on an inexpensive FR4, therefore, they are suitable for mass productions. This suggests that the proposed antennas are well suited for WiFi as well as

Akkaraekthalin, P.; Chaimool, S.; Krairiksk, M. (September 2007) Wideband uni-directional

Chaimool, S.; Akkaraekthalin P.; Krairiksh, M.(May 2011). Wideband Constant beamwidth

*Computer – Aided Engineering*, vol. 21, no 3, pp. 263-271, ISSN 1099-047X Chaimool, S.; Akkaraekthalin, P.; Vivek, V. (December 2005). Dual-band CPW-fed slot

*Transactions on Electronics,* vol. E88-C, no.12, pp.2258-2265, ISSN 0916-8524. Chaimool, S.; Chung, K. L. (2009). CPW-fed mirrored-L monopole antenna with distinct

Chaimool, S.; Jirasakulporn, P.; Akkaraekthalin, P. (2008) A new compact dual-band CPW-

Chaimool, S.; Kerdsumang, S.; Akkraeakthalin, P.; Vivek, V.(2004) A broadband CPW-fed

Jirasakulporn, P. (December 2008). Multiband CPW-fed slot antenna with L-slot bowtie

Mahatthanajatuphat, C. ; Akkaraekthalin, P.; Saleekaw, S.; Krairiksh, M. (2009). A

Moeikham, P.; Mahatthanajatuphat, C.; Akkaraekthalin, P.(2011). A compact ultrawideband

*Electromagnetics Research*, vol. 95, pp. 59-72, ISSN 1070-4698

May 17-19, 2011, ISBN: 978-1-4577-0425-3

*Technologies*, Sapporo, Japan, vol. 2, pp. 730-733, ISBN: 0-7803-8593-4 Hongnara, T.; Mahatthanajatuphat C.; Akkaraekthalin, P. (2011). Study of CPW-fed slot

CPW-fed slot antennas using loading metallic strips and a widened tuning stub on modified-shape reflectors, *IEICE Trans Communications,* vol. E90-B, no.9, pp.2246-

coplanar waveguide-fed slot antennas using metallic strip loading and a wideband tuning stub with shaped reflector, *International Journal of RF and Microwave* 

antennas using loading metallic strips and a widened tuning stub, *IEICE* 

triple bands for WiFi and WiMAX applications, *Electronics Letters,* vol. 45, no. 18,

fed slot antenna with inverted-F tuning stub, *Proceedings of ISAP-2008 International Symposium on Antennas and Propagation*, Taipei, Taiwan, pp. 1190-1193, ISBN: 978-4-

square slot antenna using loading metallic strips and a widened tuning stub, *Proceedings of ISCIT 2004 International Symposium on Communications and Information* 

antennas with fractal stubs, *Proceedings of ECTI-CON2011 8th International Conference of Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology,* pp. 188-191, Khonkean, Thailand, May 17-19, 2011, ISBN: 978-1-4577-

tuning stub, *World Academy of Science, Engineering and Technology*, vol. 48, pp.72-76,

bidirectional multiband antenna with modified fractal slot fed by CPW, *Progress In* 

monopole antenna with tapered CPW feed and slot stubs, *Proceedings of ECTI-CON2011 8th International Conference of Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology*, pp. 180-183, Khonkean, Thailand,

WiMAX portable units and base stations.

2255, ISSN 0916-8516.

pp. 928-929, ISSN 0916-8524.

88552-223-9

0425-3

ISSN 2010-376X

**7. References** 

Fig. 36. Measured radiation patterns for the CPW-fed slot antenna with the metasurface in *H-*plane. (a) 2400 MHz, (b) 2450 MHz and (c) 2500 MHz

Fig. 37. Simulated and measured realized gains of the CPW-fed slot antenna with the metasurface

#### **6. Conclusions**

In this chapter, we have introduced wideband CPW-fed slot antennas, multiband CPW-fed slot and monopole antennas, and unidirectional CPW-fed slot antennas. For multiband operation, CPW-fed multi-slots and multiple monopoles are presented. In addition to, the CPW-fed slot antenna with fractal tuning stub is also obtained for multiband operations. Some WiFi or WiMAX applications such as point-to-point communications require the unidirectional antennas. Therefore, we also present the CPW-fed slot antennas with unidirectional radiation patterns by using modified reflector and metasurface. Moreover, all of antennas are fabricated on an inexpensive FR4, therefore, they are suitable for mass productions. This suggests that the proposed antennas are well suited for WiFi as well as WiMAX portable units and base stations.

#### **7. References**

46 Advanced Transmission Techniques in WiMAX

(a) (b) (c) Fig. 36. Measured radiation patterns for the CPW-fed slot antenna with the metasurface in

Fig. 37. Simulated and measured realized gains of the CPW-fed slot antenna with the

In this chapter, we have introduced wideband CPW-fed slot antennas, multiband CPW-fed slot and monopole antennas, and unidirectional CPW-fed slot antennas. For multiband operation, CPW-fed multi-slots and multiple monopoles are presented. In addition to, the CPW-fed slot antenna with fractal tuning stub is also obtained for multiband operations. Some WiFi or WiMAX applications such as point-to-point communications require the unidirectional antennas. Therefore, we also present the CPW-fed slot antennas with unidirectional radiation patterns by using modified reflector and metasurface. Moreover, all

*H-*plane. (a) 2400 MHz, (b) 2450 MHz and (c) 2500 MHz

metasurface

**6. Conclusions** 


**3** 

**A Reconfigurable Radial Line** 

*School of Computer and Communication Engineering,* 

*University of Malaysia Perlis (UniMAP) to University Malaysia Perlis,* 

WiMAX refers to interoperable deployments of IEEE 802.16 protocol, in similarity with wireless fidelity (Wi-Fi) of IEEE 802.11 protocol but providing a larger radius of coverage. WiMAX is a potential replacement for current mobile technologies such as Global System Mobile (GSM) and High Speed Downlink Packet Access (HSDPA) and can be also applied

WiMAX is a broadband platform and needs larger bandwidth compared to existing cellular bandwidth. Fixed WiMAX used fiber optic networks instead of copper wire which is deployed in other technology. WiMAX has been successfully provided three up to four times performance of current 3G technology, and ten times performance is expected in the future. Currently, the operating frequencies of WiMAX are at 2.3 GHz, 2.5 GHz, and 3.5 GHz whereas the chip of WiMAX that operated in those frequencies is already integrated into the laptops and netbooks. As transmitter, TELCO Company requires to prepare a better transmitting communication tower in providing better WiMAX's coverage and data rates. Hence, the need of superior reconfigurable WiMAX's antenna is extremely crucial to sustain

Traditional transmission line microstrip antenna has been widely used as a reconfigurable antenna due to its less complexity and easiness to fabricate. However, the reconfigurable beam shape application especially point-to-point communication required an antenna that can provide a better gain since incorporating a PIN diode switches has been known to deteriorate the gain characteristic of an antenna [1, 7]. A lot of efforts have been allocated to enhance the gain of the conventional microstrip antenna [2-3, 5, 9]. For high gain purpose, a radial line slot array (RLSA) antenna design is more beneficial [5]. An RLSA antenna has as much as 50% higher gain than the conventional microstrip antenna [6]. Conventionally, the RLSA antenna has no reconfigurable ability due to its feeding structure which is via coaxialto-waveguide transition probe. However, it is made realizable by using feed line, PIN

**1. Introduction** 

as overlay in order to enlarge the capacity and speed.

the signal strength at the highest level (dB).

diodes and an aperture coupled feeding structure [7-8, 10-12].

**Slot Array Antenna for** 

**WiMAX Application** 

Mohd Faizal Jamlos

 *Kangar, Perlis, Malaysia* 


### **A Reconfigurable Radial Line Slot Array Antenna for WiMAX Application**

Mohd Faizal Jamlos

*School of Computer and Communication Engineering, University of Malaysia Perlis (UniMAP) to University Malaysia Perlis, Kangar, Perlis, Malaysia* 

#### **1. Introduction**

48 Advanced Transmission Techniques in WiMAX

Rakluea, C.; Chaimool, S.; Akkaraekthalin, P. (2011). Unidirectional CPW-fed slot antenna

Sari-Kha, K.; Vivek, V.; Akkaraekthalin, P. (2006) A broadband CPW-fed equilateral

October 18-20, 2006, ISBN 0-7803-9741-X.

0425-3

using metasurface, *Proceedings of ECTI-CON2011 8th International Conference of Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology*, pp. 184-187, Khonkean, Thailand, May 17-19, 2011, ISBN: 978-1-4577-

hexagonal slot antenna, *Proceedings of ISCIT 2006 International Symposium on Communications and Information Technologies*, Bangkok, Thailand, pp. 783-786,

> WiMAX refers to interoperable deployments of IEEE 802.16 protocol, in similarity with wireless fidelity (Wi-Fi) of IEEE 802.11 protocol but providing a larger radius of coverage. WiMAX is a potential replacement for current mobile technologies such as Global System Mobile (GSM) and High Speed Downlink Packet Access (HSDPA) and can be also applied as overlay in order to enlarge the capacity and speed.

> WiMAX is a broadband platform and needs larger bandwidth compared to existing cellular bandwidth. Fixed WiMAX used fiber optic networks instead of copper wire which is deployed in other technology. WiMAX has been successfully provided three up to four times performance of current 3G technology, and ten times performance is expected in the future. Currently, the operating frequencies of WiMAX are at 2.3 GHz, 2.5 GHz, and 3.5 GHz whereas the chip of WiMAX that operated in those frequencies is already integrated into the laptops and netbooks. As transmitter, TELCO Company requires to prepare a better transmitting communication tower in providing better WiMAX's coverage and data rates. Hence, the need of superior reconfigurable WiMAX's antenna is extremely crucial to sustain the signal strength at the highest level (dB).

> Traditional transmission line microstrip antenna has been widely used as a reconfigurable antenna due to its less complexity and easiness to fabricate. However, the reconfigurable beam shape application especially point-to-point communication required an antenna that can provide a better gain since incorporating a PIN diode switches has been known to deteriorate the gain characteristic of an antenna [1, 7]. A lot of efforts have been allocated to enhance the gain of the conventional microstrip antenna [2-3, 5, 9]. For high gain purpose, a radial line slot array (RLSA) antenna design is more beneficial [5]. An RLSA antenna has as much as 50% higher gain than the conventional microstrip antenna [6]. Conventionally, the RLSA antenna has no reconfigurable ability due to its feeding structure which is via coaxialto-waveguide transition probe. However, it is made realizable by using feed line, PIN diodes and an aperture coupled feeding structure [7-8, 10-12].

A Reconfigurable Radial Line Slot Array Antenna for WiMAX Application 51

(a) (b)

(c) (d)

of the AC and block the direct current (DC) simultaneously.

(c) Alignment of aperture slots and feed line (d) RLSA radiating surface

Fig. 1. Simulation structure of the proposed antenna (a) feed line (b) Aperture slots

Figure 2 shows the photographs of the proposed antenna. Each of the PIN diodes is surrounded by two inductors and two capacitors forming the switching circuit as shown in figure 2(a). The inductors intend to choke off the alternating current (AC) and radio frequency (RF) signals from flowing into the feeding line while the capacitors allow the flow

The proposed antenna is developed using an aperture coupled configuration where the upper and bottom substrate are made of FR4 dielectric substrates (relative permittivity = 4.7, loss tangent = 0.019). The sizes of the substrates are 150 mm x 150 mm. The feed probe's radius is 0.5 mm while the heights of the substrates are 1.6 mm. The back plane reflector is a made up of copper foil with 0.035 mm thickness. The foil is attached on a piece of 2 mm thickness wood. The reflector is placed under the proposed antenna by using PCB stands of 5 mm height, as shown in figure 2(d). The height between the reflector and the feed line is influential in determining the operating frequency of the antenna. If the height is larger than

Another significant problem of conventional microstrip antenna is the narrowing of halfpower beamwidth (HPBW) which could only cover forward radiated beam from −50◦ to 50◦ [9]. This antenna also has another salient advantage where it can generate a broadside radiation pattern with a wider HPBW covering from −85° to 85*°*. Such wide HPBW is deemed as an interesting characteristic in which the antenna can function as WiMAX application.

As the proposed antenna is etched from FR4 substrate, it is inexpensive in terms of fabrication. Dimension wise, the proposed antenna length and width are 150 mm and 150 mm respectively, which is smaller than conventional microstrip antenna that could achieve the same function and performance [10]. In [3, 8, 9-13], switching mechanisms are utilized to alter the radiation pattern efficiently. The antenna, proposed in this paper, can dynamically be used in a beam shaping and broadside radiation pattern for WiMAX application.

This chapter is organized as follows: In Section 2, the RLSA radiating surface, aperture slots and feed line designs incorporates with PIN diode switches are explained and the effects of different configuration of the switches are investigated. The measurement and simulation of beam shaping and broadside radiation pattern using PIN diodes switching results will be shown in Section 3. Finally, conclusion will be drawn in Section 4.

#### **2. Antenna structure**

The proposed antenna structure, as shown in figure 1, has the ability to exhibit two major types of radiation patterns; the beam shape and the broadside radiation pattern. The 'circular' and a 'bridge' feed line are interconnected by switches, which consists of end-fire beam-shaped reconfigurable switches (EBRS) and broadside reconfigurable switches (BRS). The EBRS are referring to the first up to the fifth switches while the BRS are the first, fifth, sixth and seventh switches as shown in figure 1(a).

Four aperture slots are used to couple the feeding line to the radiating surface as shown in figure 1(b). Inaccuracy of alignment between the layer of feed line and aperture slots to the radiating surface can significantly deteriorate the antenna's performance especially on the gain characteristic. The aperture slots determine the amount of coupling to the RLSA radiating surface from the feed line of the proposed antenna. Hence, the feed line must be aligned beneath the aperture slots accurately as shown in figure 1(c). The length of the four aperture slots are 40 mm while their width are 3 mm.

The RLSA pattern that is used as the radiating surface in the proposed antenna has the arrangement as shown in figure 1(d) in order to provide a linear polarization along the beam direction. There are 96 slots, with 16 slots in the inner-most ring, and 32 slots in the outer-most ring. The width and length of the RLSA slots are 1.5 mm and 15 mm respectively. The gaps between the slots are mostly 8 mm. The diameter of the circular radiating surface is 150 mm.

Generally, by turning the EBRS ON and the sixth and seventh of the BRS OFF, it will result in a beam shape radiation pattern. The pattern will becomes narrower with an increasing number of EBRS switches turned ON. While by turning ON the BRS and the second up to fourth of EBRS turned OFF, a broadside radiation pattern will be obtained.

Another significant problem of conventional microstrip antenna is the narrowing of halfpower beamwidth (HPBW) which could only cover forward radiated beam from −50◦ to 50◦ [9]. This antenna also has another salient advantage where it can generate a broadside radiation pattern with a wider HPBW covering from −85° to 85*°*. Such wide HPBW is deemed as an interesting characteristic in which the antenna can function as WiMAX

As the proposed antenna is etched from FR4 substrate, it is inexpensive in terms of fabrication. Dimension wise, the proposed antenna length and width are 150 mm and 150 mm respectively, which is smaller than conventional microstrip antenna that could achieve the same function and performance [10]. In [3, 8, 9-13], switching mechanisms are utilized to alter the radiation pattern efficiently. The antenna, proposed in this paper, can dynamically

This chapter is organized as follows: In Section 2, the RLSA radiating surface, aperture slots and feed line designs incorporates with PIN diode switches are explained and the effects of different configuration of the switches are investigated. The measurement and simulation of beam shaping and broadside radiation pattern using PIN diodes switching results will be

The proposed antenna structure, as shown in figure 1, has the ability to exhibit two major types of radiation patterns; the beam shape and the broadside radiation pattern. The 'circular' and a 'bridge' feed line are interconnected by switches, which consists of end-fire beam-shaped reconfigurable switches (EBRS) and broadside reconfigurable switches (BRS). The EBRS are referring to the first up to the fifth switches while the BRS are the first, fifth,

Four aperture slots are used to couple the feeding line to the radiating surface as shown in figure 1(b). Inaccuracy of alignment between the layer of feed line and aperture slots to the radiating surface can significantly deteriorate the antenna's performance especially on the gain characteristic. The aperture slots determine the amount of coupling to the RLSA radiating surface from the feed line of the proposed antenna. Hence, the feed line must be aligned beneath the aperture slots accurately as shown in figure 1(c). The length of the four

The RLSA pattern that is used as the radiating surface in the proposed antenna has the arrangement as shown in figure 1(d) in order to provide a linear polarization along the beam direction. There are 96 slots, with 16 slots in the inner-most ring, and 32 slots in the outer-most ring. The width and length of the RLSA slots are 1.5 mm and 15 mm respectively. The gaps between the slots are mostly 8 mm. The diameter of the circular

Generally, by turning the EBRS ON and the sixth and seventh of the BRS OFF, it will result in a beam shape radiation pattern. The pattern will becomes narrower with an increasing number of EBRS switches turned ON. While by turning ON the BRS and the second up to

fourth of EBRS turned OFF, a broadside radiation pattern will be obtained.

be used in a beam shaping and broadside radiation pattern for WiMAX application.

shown in Section 3. Finally, conclusion will be drawn in Section 4.

sixth and seventh switches as shown in figure 1(a).

aperture slots are 40 mm while their width are 3 mm.

application.

**2. Antenna structure** 

radiating surface is 150 mm.

Fig. 1. Simulation structure of the proposed antenna (a) feed line (b) Aperture slots (c) Alignment of aperture slots and feed line (d) RLSA radiating surface

Figure 2 shows the photographs of the proposed antenna. Each of the PIN diodes is surrounded by two inductors and two capacitors forming the switching circuit as shown in figure 2(a). The inductors intend to choke off the alternating current (AC) and radio frequency (RF) signals from flowing into the feeding line while the capacitors allow the flow of the AC and block the direct current (DC) simultaneously.

The proposed antenna is developed using an aperture coupled configuration where the upper and bottom substrate are made of FR4 dielectric substrates (relative permittivity = 4.7, loss tangent = 0.019). The sizes of the substrates are 150 mm x 150 mm. The feed probe's radius is 0.5 mm while the heights of the substrates are 1.6 mm. The back plane reflector is a made up of copper foil with 0.035 mm thickness. The foil is attached on a piece of 2 mm thickness wood. The reflector is placed under the proposed antenna by using PCB stands of 5 mm height, as shown in figure 2(d). The height between the reflector and the feed line is influential in determining the operating frequency of the antenna. If the height is larger than

A Reconfigurable Radial Line Slot Array Antenna for WiMAX Application 53

**dB dB** 

(a) (b)

**dB** 

(c) (d)

Fig. 3. The measurement of beam shape radiation patterns by turning ON the EBRS (a) switch i (b) switches i and ii (c) switches i, ii, and iii (d) switches i, ii, iii, iv and v

In figure 3(d), the HPBW of the radiation pattern is from -10° to 15° and antenna gain of 14.64 dB are obtained when the first until the fifth switches of the EBRS are turned ON. It is obvious that the proposed antenna can be tuned to have a wide HPBW which covered from -65° to 70° beam angle range, compared to conventional microstrip antenna that can only cover from -50° to 50° beam angle range. The maximum gain of the proposed antenna is 14.64 dB, which can be considered high for an antenna of such size. Computer Simulation Technology (CST) Studio Suite 2009 is used as a platform to design and simulate the radiation pattern of the proposed antenna. It is clearly shown that the simulations have the same behaviour with the measurements where the higher the produced gain, the antenna's

**dB** 

its optimized height, which is 5 mm for this antenna, the operating frequency will be shifted to a lower centre frequency, and vice versa. The reflector width and length are both 150 mm, thus making its surface area the same as the size of the antenna. The proposed antenna is operating at frequency of 2.3 GHz.

Fig. 2. Photograph of the proposed antenna (a) Feed line with PIN diodes switches (b) Aperture slots (c) RLSA radiating surface (d) side view (e) Layout view

#### **3. Result and discussion**

Measurement shows that four different types of beam shape radiation pattern can be well reconfigured with the configuration of the EBRS. Different activation of EBRS will results in different gain and HPBW. By turning ON the first switch of the EBRS, gain and HPBW of 4.85 dB and -65° to 70° are obtained respectively, as shown in figure 3(a). While in figure 3(b), turning ON the first and second switches of the EBRS will narrow the HPBW from -40° to 45° with a gain of 7.2 dB. Figure 3(c) demonstrates the beam shape of the radiation pattern with the HPBW from -15° to 20° and a gain of 9.9 dB by turning ON the first, second and third switch of the EBRS simultaneously.

its optimized height, which is 5 mm for this antenna, the operating frequency will be shifted to a lower centre frequency, and vice versa. The reflector width and length are both 150 mm, thus making its surface area the same as the size of the antenna. The proposed antenna is

(a) (b)

(c) (d) (e)

(b) Aperture slots (c) RLSA radiating surface (d) side view (e) Layout view

Fig. 2. Photograph of the proposed antenna (a) Feed line with PIN diodes switches

Measurement shows that four different types of beam shape radiation pattern can be well reconfigured with the configuration of the EBRS. Different activation of EBRS will results in different gain and HPBW. By turning ON the first switch of the EBRS, gain and HPBW of 4.85 dB and -65° to 70° are obtained respectively, as shown in figure 3(a). While in figure 3(b), turning ON the first and second switches of the EBRS will narrow the HPBW from -40° to 45° with a gain of 7.2 dB. Figure 3(c) demonstrates the beam shape of the radiation pattern with the HPBW from -15° to 20° and a gain of 9.9 dB by turning ON the first, second

operating at frequency of 2.3 GHz.

**3. Result and discussion** 

and third switch of the EBRS simultaneously.

Fig. 3. The measurement of beam shape radiation patterns by turning ON the EBRS (a) switch i (b) switches i and ii (c) switches i, ii, and iii (d) switches i, ii, iii, iv and v

In figure 3(d), the HPBW of the radiation pattern is from -10° to 15° and antenna gain of 14.64 dB are obtained when the first until the fifth switches of the EBRS are turned ON. It is obvious that the proposed antenna can be tuned to have a wide HPBW which covered from -65° to 70° beam angle range, compared to conventional microstrip antenna that can only cover from -50° to 50° beam angle range. The maximum gain of the proposed antenna is 14.64 dB, which can be considered high for an antenna of such size. Computer Simulation Technology (CST) Studio Suite 2009 is used as a platform to design and simulate the radiation pattern of the proposed antenna. It is clearly shown that the simulations have the same behaviour with the measurements where the higher the produced gain, the antenna's

A Reconfigurable Radial Line Slot Array Antenna for WiMAX Application 55

achieves a higher gain in comparison to the divisive broadside pattern. Figure 8 depicts the 3D simulation of the far field radiation patterns of the proposed antenna which is aligned with the measured radiation patterns as shown in figure 6 and figure 7. However, the measured gain and HPBW are slightly less compared to the simulations due to CST

simulation's ideal and free loss environment.

dB

 (a) (b) Fig. 6. Broadside divisive radiation patterns (a) Measured (b) Simulation

dB

(a) (b)

Fig. 7. Broadside single-sided radiation patterns (a) Measured (b) Simulation

HPBW will becomes narrower as shown in figure 4(a) and figure 4(b). 3D representation of the far field radiation patterns are shown in figure 5. The measurement has comparable results with the simulations. Nevertheless, the measured antenna's gain is slightly higher than simulation results but smaller in terms of HPBW.

Fig. 4. The complete beam shape radiation patterns (a) Measured (b) Simulation

Fig. 5. 3D-polar plot of beam shape radiation patterns by turning ON the EBRS (a) switch i (b) switches i and ii (c) switches i, ii, and iii (d) switches i, ii, iii, iv and v

The BRS configuration has the ability to turn the radiation pattern from the beam shape to broadside radiation pattern perfectly as shown in figure 6 and figure 7. Figure 6 shows a divisive broadside radiation pattern with a maximum gain of 10.8 dB and a wider HPBW of -85°- 85° when turning all of the BRS ON simultaneously. Certain combination configuration between the BRS and EBRS are able to generate another radiation pattern which is a broadside single-sided radiation pattern. This kind of pattern is lean to the right with HPBW of -80°- 80° when turning ON the sixth and seventh of BRS and the first up to fourth of EBRS concurrently with a maximum gain increased up to 12.8 dB as shown in Figure 7. Since in this switching configuration the direction of radiation pattern is focused on one side, it

HPBW will becomes narrower as shown in figure 4(a) and figure 4(b). 3D representation of the far field radiation patterns are shown in figure 5. The measurement has comparable results with the simulations. Nevertheless, the measured antenna's gain is slightly higher

(a) (b)

Fig. 4. The complete beam shape radiation patterns (a) Measured (b) Simulation

(a) (b) (c) (d)

(b) switches i and ii (c) switches i, ii, and iii (d) switches i, ii, iii, iv and v

Fig. 5. 3D-polar plot of beam shape radiation patterns by turning ON the EBRS (a) switch i

The BRS configuration has the ability to turn the radiation pattern from the beam shape to broadside radiation pattern perfectly as shown in figure 6 and figure 7. Figure 6 shows a divisive broadside radiation pattern with a maximum gain of 10.8 dB and a wider HPBW of -85°- 85° when turning all of the BRS ON simultaneously. Certain combination configuration between the BRS and EBRS are able to generate another radiation pattern which is a broadside single-sided radiation pattern. This kind of pattern is lean to the right with HPBW of -80°- 80° when turning ON the sixth and seventh of BRS and the first up to fourth of EBRS concurrently with a maximum gain increased up to 12.8 dB as shown in Figure 7. Since in this switching configuration the direction of radiation pattern is focused on one side, it

than simulation results but smaller in terms of HPBW.

**d**

achieves a higher gain in comparison to the divisive broadside pattern. Figure 8 depicts the 3D simulation of the far field radiation patterns of the proposed antenna which is aligned with the measured radiation patterns as shown in figure 6 and figure 7. However, the measured gain and HPBW are slightly less compared to the simulations due to CST simulation's ideal and free loss environment.

Fig. 6. Broadside divisive radiation patterns (a) Measured (b) Simulation

Fig. 7. Broadside single-sided radiation patterns (a) Measured (b) Simulation

A Reconfigurable Radial Line Slot Array Antenna for WiMAX Application 57

PIN diode status

i ON ON ON ON ON ON

ii OFF ON ON ON OFF ON

iii OFF OFF ON ON OFF ON

iv OFF OFF OFF ON OFF ON

v OFF OFF OFF ON ON OFF

vi OFF OFF OFF OFF ON ON

vii OFF OFF OFF OFF ON ON

**Divisive radiation pattern** 

**Single sided radiation pattern** 

Gain (dB) **4.85 7.2 9.9 14.64 10.8 12.8** 

HPBW(°) **-65°- 70° -40° to 45° -15° to 20° -10° to 15° -85°- 85° -80°- 80°** 

Table 1. Configuration of PIN diode switches of the measured proposed antenna at 2.3 GHz

A novel reconfigurable radiation pattern microstrip antenna using RLSA is introduced in this paper. This chapter has taken advantages of the high performances of RLSA in terms of gain and less signals reflection, to make the proposed antenna becomes more efficient. This antenna is designed based on aperture coupled structure. The ability of the beam shape and broadside radiation pattern is attributed with the usage of PIN diode switches that integrated in the feed line of the proposed antenna. It is shown through the measurements that the radiation patterns can be well reconfigured through the assist of the orientation and geometry of the RLSA slots. The proposed antenna which has a dimension size of 150 mm X 150 mm, can be tuned to reach a high gain of 14.64 dB. The antenna can also provide wider value of HPBW that covered from -85° to 85° which is far better than -50° to 50° HPBW of a conventional microstrip antenna. The broadside patterns are achieved by turning ON selected configuration of the combination between the BRS and EBRS. The structure of the proposed antenna which is not bulky compared to the conventional microstrip antenna would be greatly suitable for beam shape and broadside radiation pattern application such

Type of switch

End-fire beamshaped reconfigura ble switches (EBRS)

Broadside reconfigura ble switches (BRS)

Type of radiation

**4. Conclusion** 

as WiMAX.

pattern **Beam shaping** 

Number of PIN diode switch

Fig. 8. 3D-polar plot of broadside radiation patterns (a) Divisive (b) Single-sided

The measurements show a very good agreement with simulations where the radiation patterns are formed successfully with respect to the beam shaping and broadside characteristics. The high gain measurements and simulations of the proposed antenna can be attributed to the good coupling from the feed line to the RLSA radiating surface through the appropriate sizing, positioning and shape of aperture slots. The outputs of the PIN diodes switching scheme involving the EBRS and BRS are summarized in table 1. All the radiation patterns of the proposed antenna are relatively at frequency 2.3 GHz as depicted by figure 9.

Fig. 9. The measurement of return loss by variation activation of the EBRS and BRS


Table 1. Configuration of PIN diode switches of the measured proposed antenna at 2.3 GHz

#### **4. Conclusion**

56 Advanced Transmission Techniques in WiMAX

(a) (b)

by figure 9.

Fig. 8. 3D-polar plot of broadside radiation patterns (a) Divisive (b) Single-sided

Fig. 9. The measurement of return loss by variation activation of the EBRS and BRS

The measurements show a very good agreement with simulations where the radiation patterns are formed successfully with respect to the beam shaping and broadside characteristics. The high gain measurements and simulations of the proposed antenna can be attributed to the good coupling from the feed line to the RLSA radiating surface through the appropriate sizing, positioning and shape of aperture slots. The outputs of the PIN diodes switching scheme involving the EBRS and BRS are summarized in table 1. All the radiation patterns of the proposed antenna are relatively at frequency 2.3 GHz as depicted

> A novel reconfigurable radiation pattern microstrip antenna using RLSA is introduced in this paper. This chapter has taken advantages of the high performances of RLSA in terms of gain and less signals reflection, to make the proposed antenna becomes more efficient. This antenna is designed based on aperture coupled structure. The ability of the beam shape and broadside radiation pattern is attributed with the usage of PIN diode switches that integrated in the feed line of the proposed antenna. It is shown through the measurements that the radiation patterns can be well reconfigured through the assist of the orientation and geometry of the RLSA slots. The proposed antenna which has a dimension size of 150 mm X 150 mm, can be tuned to reach a high gain of 14.64 dB. The antenna can also provide wider value of HPBW that covered from -85° to 85° which is far better than -50° to 50° HPBW of a conventional microstrip antenna. The broadside patterns are achieved by turning ON selected configuration of the combination between the BRS and EBRS. The structure of the proposed antenna which is not bulky compared to the conventional microstrip antenna would be greatly suitable for beam shape and broadside radiation pattern application such as WiMAX.

**0**

**4**

**Reduction of Nonlinear Distortion in**

Multiple-input multiple-output (MIMO) techniques in combination with orthogonal frequency-division multiplexing (OFDM) have already found its deployment in several standards for the broadband communications including WiMAX or 3GPP proposal termed as Long Term Evolution (LTE). The MIMO-OFDM allows to substantially increase the spectral efficiency, link reliability and coverage of the signal transmission. With recent advent of the hardware processing enhacements, the processing requirements of MIMO-OFDM might be accomodated in the portable units and thus, it is widely expected that this technology will

Despite of its undoubted benefits, MIMO-OFDM transmission systems are also characterized by the large envelope fluctuation of the transmitted signal Drotar et al. (2010a). This requires the application of the high Input Back-off (IBO) at the nonlinear High Power Amplifier (HPA) stage that subsequently results in an inefficient use of HPA and limitation of the battery life in

In is important to note that nonlinear amplification manifests itself in the form of Bit-Error-Rate (BER) degradation at the receiver side and simultaneously, in the form of the out-of-band radiation Deumal et al. (2008). An intuitive solution to supress the out-of-band radiation and thus, occupy the area within the spectral mask of the transmission is is to deactivate subcarriers at the borders of the used MIMO-OFDM spectrum. However, this approach impairs the spectral efficiency of the transmission and may not be convenient for the high data rate applications. Therefore it is feasible to look for the additional technique that aim to reduce the out-of-band emissions and to maintain the specific spectral mask of the

The possible solution is to design MIMO-OFDM systems such that the signal is less sensitive to the nonlinearity impairments. Lower fluctuation of the signal envelope can be achieved by modifying the transmitted signal prior to the transmission. However, this approach requires additional hardware and signal processing at the transmitter, which is not feasible in some applications. For these applications, the receiver based compensation is of more interest. In the following sections, we will review the details of the most favourite methods reducing envelope fluctuation, which are intended to be used in Single-Input Single-Output (SISO) OFDM and MIMO-OFDM systems. Moreover, we will introduce two novel techniques that

dominate over the next years in the wireless communications.

**1. Introduction**

the user mobile stations.

transmission Khan (2009).

**Multi-Antenna WiMAX Systems**

Peter Drotár, Juraj Gazda, Dušan Kocur

and Pavol Galajda

*Slovakia*

*Technical University of Kosice*

#### **5. References**


### **Reduction of Nonlinear Distortion in Multi-Antenna WiMAX Systems**

Peter Drotár, Juraj Gazda, Dušan Kocur and Pavol Galajda *Technical University of Kosice Slovakia*

#### **1. Introduction**

58 Advanced Transmission Techniques in WiMAX

[1] G. Monti, R. De Paolis, and L. Tarricone, "Design of a 3-state reconfigurable crlh

[2] M. T. Islam, M. N. Shakib and N. Misran, "Design Analysis Of High Gain Wideband L-

[3] X. Li, L. Yang, S.-X. Gong, and Y.-J. Yang, "Bidirectional High Gain Antenna for Wlan Applications," *Progress In Electromagnetics Research Letters*, Vol. 6, 99-106, 2009. [4] N. Romano, G. Prisco and F. Soldovieri, "Design Of A Reconfigurable Antenna For

[5] O. Beheshti-Zavareh and M. Hakak, "A Stable Design Of Coaxial Adaptor For Radial Line Slot Antenna", *Progress In Electromagnetics Research,* PIER 90, 51–62, 2009. [6] Paul W. Davis and Marek E. Bialkowski, "Experimental Investigations into a Linearly

[7] J. L. Masa-Campos and F. Gonzalez-Fernandez, "Dual linear/circular polarized plannar

[9] J. Ouyang, F. Yang, S. W. Yang, Z. P. Nie and Z. Q. Zhao., "A novel radiation pattern and

[10] M. F. Jamlos, O. A. Aziz, T. A. Rahman and M. R. Kamarudin "a beam steering radial

[11] M. F. Jamlos, T. A. Rahman, and M. R. Kamarudin "a novel adaptive wi-fi system with rfid Technology" *Progress In Electromagnetics Research, Vol. 108, 417-432, 2010* [12] M. F. Jamlos, T. A. Rahman, and M. R. Kamarudin "adaptive beam steering of rlsa

[13] Huff, G. H., J. Feng, S. Zhang, and J. T. Bernhard, "A novel radiation pattern and

*Microwave Wireless Components Letter*, Vol. 13, 57–59, February 2003.

Electromagn. Waves and Appl., Vol. 24, 1079–1088, 2010

*Antennas and Propagation*, vol. 45, No. 7, July1997.

transmission line based on mems switches," *Progress In Electromagnetics Research*,

Probe Fed Microstrip Patch Antenna" *Progress In Electromagnetics Research,* PIER 95,

Ground Penetrating Radar Applications" *Progress In Electromagnetics Research,* PIER

Polarized Radial Slot Antenna for DBS TV in Australia", *IEEE Transactions on* 

antenna with low profile double-layer polarizer of 45º tilted metallic strips for wimax applications," *Progress In Electromagnetics Research*, PIER 98, 221-231, 2009. [8] G. M. Rebeiz, RF MEMS: Theory, Design, and Technology. Hoboken, NJ: Wiley-

frequency reconfigurable microstrip antenna on a thin substrate for wide-band and wide-angle scanning application", *Progress In Electromagnetics Research*, PIER 4, 167–

line slot array (rlsa) antenna with reconfigurable operating frequency" J. of

antenna with rfid technology," *Progress In Electromagnetics Research, Vol. 108, 65-80,* 

frequency reconfigurable single turn square spiral microstrip antenna," *IEEE* 

**5. References** 

PIER 95, 283-297, 2009.

397-407, 2009.

94, 1-18, 2009.

Interscience, 2003.

172, 2008.

*2010* 

Multiple-input multiple-output (MIMO) techniques in combination with orthogonal frequency-division multiplexing (OFDM) have already found its deployment in several standards for the broadband communications including WiMAX or 3GPP proposal termed as Long Term Evolution (LTE). The MIMO-OFDM allows to substantially increase the spectral efficiency, link reliability and coverage of the signal transmission. With recent advent of the hardware processing enhacements, the processing requirements of MIMO-OFDM might be accomodated in the portable units and thus, it is widely expected that this technology will dominate over the next years in the wireless communications.

Despite of its undoubted benefits, MIMO-OFDM transmission systems are also characterized by the large envelope fluctuation of the transmitted signal Drotar et al. (2010a). This requires the application of the high Input Back-off (IBO) at the nonlinear High Power Amplifier (HPA) stage that subsequently results in an inefficient use of HPA and limitation of the battery life in the user mobile stations.

In is important to note that nonlinear amplification manifests itself in the form of Bit-Error-Rate (BER) degradation at the receiver side and simultaneously, in the form of the out-of-band radiation Deumal et al. (2008). An intuitive solution to supress the out-of-band radiation and thus, occupy the area within the spectral mask of the transmission is is to deactivate subcarriers at the borders of the used MIMO-OFDM spectrum. However, this approach impairs the spectral efficiency of the transmission and may not be convenient for the high data rate applications. Therefore it is feasible to look for the additional technique that aim to reduce the out-of-band emissions and to maintain the specific spectral mask of the transmission Khan (2009).

The possible solution is to design MIMO-OFDM systems such that the signal is less sensitive to the nonlinearity impairments. Lower fluctuation of the signal envelope can be achieved by modifying the transmitted signal prior to the transmission. However, this approach requires additional hardware and signal processing at the transmitter, which is not feasible in some applications. For these applications, the receiver based compensation is of more interest.

In the following sections, we will review the details of the most favourite methods reducing envelope fluctuation, which are intended to be used in Single-Input Single-Output (SISO) OFDM and MIMO-OFDM systems. Moreover, we will introduce two novel techniques that

−2 −1.5 −1 −0.5 0 0.5 1 1.5 2

Reduction of Nonlinear Distortion in Multi-Antenna WiMAX Systems 61

If there is a non-constant envelope signal (e.g OFDM signal) at the input of HPA, the nonlinear amplification might result in the significant nonlinear distortion that consequently affects the system performance. The resulting effect of the nonlinear distortion can be divided into the two types: the out-of-band distortion and the in-band distortion. The in-band distortion produces inter-carrier interference increasing BER, or equivalently reducing the system capacity or operational range. The out-of-band distortion appears as the spectral

The spectral regrowth can be easily explained by the intermodulation product introduced by the nonlinearity. Intermodulation products may potentially lay outside the transmission bandwidth, what means that some portion of energy is generated into the neighbouring channel. However, these channels are usually occupied by the adjacent user and so the operation point of HPA has to be chosen very carefully to meet the spectral mask constrains. Employing higher IBO values leads to the suppression of the out-of-band radiation, but at the cost of reduced HPA efficiency. Figure 1 shows the PSD curves for the OFDM signal employing *Nc* = 256 subcarriers and soft limiter model of HPA at various IBO levels. As can be seen from the figure, there is a significant out of band radiation at low IBO levels, but it decreases towards larger IBO. As the result, by applying larger IBO, HPA operates in the linear region of its characteristic. The spectral regrowth and out-of-band distortion is treated

Next, the BER performance degradation caused by the nonlinear amplification is considered. In the following we assume that the distortion caused by the HPA can be modelled as an additive Gaussian noise (AWGN) whose variance depends on the input signal and the nonlinear HPA characteristics. Note that, even though this is the most common assumption in the literature Dardari et al. (2000); Ochiai & Imai (2001); Tellado (2000), there are some cases, e.g low number of subcarriers or low clipping levels, when this assumption is inaccurate and

Fig. 1. Power spectral density at the output of the transmitter at various IBO.

regrowth, hence causing the interference in the adjacent channels.

in more detail in e.g. Baytekin & Meyer (2005); Zhou & Raich (2004).

does not hold.

Normalized frequency to 1/Td

NL input NL output

IBO [dB]

8

−50

−40

−30

−20

PSD [dB]

−10

0

aim to supress the effects of the nonlinearities in MIMO-OFDM. The former will significantly reduce the envelope fluctuation by using the null subcarriers occuring in the transmission and the latter will improve the BER performance of MIMO-OFDM by means of the iterative detection.

Specially, the salient advantage of employing the nonlinear detector scheme in WiMAX is that, since it is implemented at the base station, it does not increase the computational complexity of the mobile terminal, thus neither increasing the cost nor reducing the battery life. On the other hand, using the null subcarriers for the envelope fluctuation reduction does not reduce the data rate, nor the spectral efficiency of the transmission and therefore its application is also vital in WiMAX.

#### **2. MIMO-OFDM system model**

Given we have *Nt* transmit antennas and *Nc* OFDM subcarriers, at each time instant *t* a block of symbols is encoded to generate space-frequency codeword. The space-frequency block-code (SFBC) codeword is then given by

$$\mathbf{X} = \begin{pmatrix} \mathbf{x}\_1^1 & \mathbf{x}\_2^1 & \cdots & \mathbf{x}\_{N\_l}^1 \\ \mathbf{x}\_1^2 & \mathbf{x}\_2^2 & \cdots & \mathbf{x}\_{N\_l}^2 \\ \vdots & \vdots & \ddots & \vdots \\ \mathbf{x}\_1^{N\_c} & \mathbf{x}\_2^{N\_c} & \cdots & \mathbf{x}\_{N\_l}^{N\_c} \end{pmatrix}^\prime \tag{1}$$

where *n*-th column is the data sequence for *n*-th transmit antenna.

Space-frequency codeword is generated by grouping subcarriers and applying space time block code only across the sub-carriers in the same group Giannakis et al. (2007); Jafarkhani (2005); Liu et al. (2002). If SFBC is designed carefully, such a grouping will not degrade the diversity gain of the proposed coding scheme. Moreover, the subcarrier grouping reduces the complexity and allows the design of code matrices per subsystem since space-frequency coding constructs **X***<sup>g</sup>* separately as in (2) instead of constructing the entire **X** as in (1).

$$\mathbf{X} = \begin{pmatrix} \mathbf{x}\_0 \\ \mathbf{x}\_1 \\ \vdots \\ \mathbf{x}\_{N\_g-1} \end{pmatrix},\tag{2}$$

where *Ng* is number of sub-blocks equal to *Ng* = *Nc*/*Ns* and *Ns* is number of the time slots required to transmit one codeword.

#### **3. Problem formulation**

The discussion in this chapter assumes the single antenna system. However, the extension to MIMO-OFDM is straightforward and will be used with advantage later in the sections.

2 Will-be-set-by-IN-TECH

aim to supress the effects of the nonlinearities in MIMO-OFDM. The former will significantly reduce the envelope fluctuation by using the null subcarriers occuring in the transmission and the latter will improve the BER performance of MIMO-OFDM by means of the iterative

Specially, the salient advantage of employing the nonlinear detector scheme in WiMAX is that, since it is implemented at the base station, it does not increase the computational complexity of the mobile terminal, thus neither increasing the cost nor reducing the battery life. On the other hand, using the null subcarriers for the envelope fluctuation reduction does not reduce the data rate, nor the spectral efficiency of the transmission and therefore its application is also

Given we have *Nt* transmit antennas and *Nc* OFDM subcarriers, at each time instant *t* a block of symbols is encoded to generate space-frequency codeword. The space-frequency

> <sup>2</sup> ··· *<sup>x</sup>*<sup>1</sup> *Nt*

⎞

⎟⎟⎟⎟⎟⎟⎟⎠

, (1)

, (2)

<sup>2</sup> ··· *<sup>x</sup>*<sup>2</sup> *Nt*

<sup>2</sup> ··· *<sup>x</sup>Nc Nt*

**X** =

where *n*-th column is the data sequence for *n*-th transmit antenna.

⎛

*x*1 <sup>1</sup> *<sup>x</sup>*<sup>1</sup>

*x*2 <sup>1</sup> *<sup>x</sup>*<sup>2</sup>

> . . . . . . ... . . .

*xNc* <sup>1</sup> *<sup>x</sup>Nc*

coding constructs **X***<sup>g</sup>* separately as in (2) instead of constructing the entire **X** as in (1).

⎛

**X**0 **X**1 . . .

⎞

⎟⎟⎟⎟⎟⎟⎟⎠

**X***Ng*−<sup>1</sup>

where *Ng* is number of sub-blocks equal to *Ng* = *Nc*/*Ns* and *Ns* is number of the time slots

The discussion in this chapter assumes the single antenna system. However, the extension to MIMO-OFDM is straightforward and will be used with advantage later in the sections.

⎜⎜⎜⎜⎜⎜⎜⎝

**X** =

Space-frequency codeword is generated by grouping subcarriers and applying space time block code only across the sub-carriers in the same group Giannakis et al. (2007); Jafarkhani (2005); Liu et al. (2002). If SFBC is designed carefully, such a grouping will not degrade the diversity gain of the proposed coding scheme. Moreover, the subcarrier grouping reduces the complexity and allows the design of code matrices per subsystem since space-frequency

⎜⎜⎜⎜⎜⎜⎜⎝

detection.

vital in WiMAX.

**2. MIMO-OFDM system model**

required to transmit one codeword.

**3. Problem formulation**

block-code (SFBC) codeword is then given by

Fig. 1. Power spectral density at the output of the transmitter at various IBO.

If there is a non-constant envelope signal (e.g OFDM signal) at the input of HPA, the nonlinear amplification might result in the significant nonlinear distortion that consequently affects the system performance. The resulting effect of the nonlinear distortion can be divided into the two types: the out-of-band distortion and the in-band distortion. The in-band distortion produces inter-carrier interference increasing BER, or equivalently reducing the system capacity or operational range. The out-of-band distortion appears as the spectral regrowth, hence causing the interference in the adjacent channels.

The spectral regrowth can be easily explained by the intermodulation product introduced by the nonlinearity. Intermodulation products may potentially lay outside the transmission bandwidth, what means that some portion of energy is generated into the neighbouring channel. However, these channels are usually occupied by the adjacent user and so the operation point of HPA has to be chosen very carefully to meet the spectral mask constrains. Employing higher IBO values leads to the suppression of the out-of-band radiation, but at the cost of reduced HPA efficiency. Figure 1 shows the PSD curves for the OFDM signal employing *Nc* = 256 subcarriers and soft limiter model of HPA at various IBO levels. As can be seen from the figure, there is a significant out of band radiation at low IBO levels, but it decreases towards larger IBO. As the result, by applying larger IBO, HPA operates in the linear region of its characteristic. The spectral regrowth and out-of-band distortion is treated in more detail in e.g. Baytekin & Meyer (2005); Zhou & Raich (2004).

Next, the BER performance degradation caused by the nonlinear amplification is considered. In the following we assume that the distortion caused by the HPA can be modelled as an additive Gaussian noise (AWGN) whose variance depends on the input signal and the nonlinear HPA characteristics. Note that, even though this is the most common assumption in the literature Dardari et al. (2000); Ochiai & Imai (2001); Tellado (2000), there are some cases, e.g low number of subcarriers or low clipping levels, when this assumption is inaccurate and does not hold.

reconstruction (IAR) in Kwon & Im (2006) is presented. Another approach for improving the performance of clipped MIMO-OFDM systems with (quasy)-OSTBC is to use the statistics of the clipping distortions to develop maximum likelihood (ML) decoding Li & Xia (2008). For the case of spatially multiplexed systems, the soft correction method of Bittner et al. (2008) is

Reduction of Nonlinear Distortion in Multi-Antenna WiMAX Systems 63

In the selected mapping technique, the transmitter generates several candidate data blocks from the original data block. Subsequently, the one with the lowest envelope fluctuation is transmitted. The candidate data blocks are generated as follows. First, *U* different phase

normally **b**<sup>1</sup> is set to be all-one vector of the length *N* in order to include also unmodified block into the set of candidate data blocks. Then candidate data blocks are generated by element-wise multiplication of the frequency-domain OFDM symbol by each **b**(*u*), IFFT is applied and the resulting block with the lowest envelope fluctuations is selected for the transmission. The information about the selected phase sequence has to be transmitted to the receiver in the form of the side information. At the receiver, the reverse operation is performed

The straight-forward implementation of conventional SISO SLM is similar to PTS. In this scheme the SISO SLM technique is applied separatelly to each of the *Nt* transmitting antennas in the MIMO-OFDM system. For each of the parallel OFDM frames, the best phase

*Concurrent* SLM approach ensures higher reliability of side information. This is achieved through the spatial diversity by transmitting the same side information on different

In Fischer & Hoch (2006), authors introduce *directed* SLM (*d*SLM) scheme that uses advantage of multiple antennas. The PAPR decreasing abilities of this method improve with increasing number of antennas, however this comes at the expense of higher number of side information

In order to improve bandwidth efficiency number of transmitted SI bits has to be decreased. Therefore *small-overhead* SLM was proposed in Hassan et al. (2009). This scheme not only improve bandwidth efficiency but achieves also substantially better BER performance

The computational complexity of SLM method is relatively high therefore there is strong need for low-complexity solutions. The promising approach that require only one FFT operation was introduced in Wang & Li (2009). It exploits the time-domain signal properties of MIMO-OFDM systems to achieve a low-complexity architecture for candidate signal

In the partial transmit sequence technique Han & Lee (2006); Muller & Huber (1997), the original data block of length *N* is partitioned into *V* disjoint subblocks, **s***<sup>v</sup>* =

weighted by a phase factor, *bv* = *ej*Φ*<sup>v</sup>* , *v* = 1, 2, . . . , *V*, for *v*th subblock. Such phase factors are

(*u*) <sup>0</sup> , *b* (*u*) <sup>1</sup> ,..., *b*

(*u*)

*<sup>v</sup>*=<sup>1</sup> **s***<sup>v</sup>* = **s**. The subcarriers in each subblock are

*<sup>N</sup>*−1), *<sup>u</sup>* <sup>=</sup> 1, . . . , *<sup>U</sup>*, where

applicable.

**4.2 The selected mapping scheme**

to recover the original data block.

antennas Lee et al. (2003).

(SI) bits.

generation.

sequences of length *N* are generated, **b**(*u*) = (*b*

modification out of the *U* possible ones is individually selected.

compared to *d*SLM or SISO SLM applied on multiple antennas.

**4.3 The partial transmit sequence technique**

(*sv*,0,*sv*,1,...,*sv*,*N*−1), *<sup>v</sup>* <sup>=</sup> 1, . . . , *<sup>V</sup>* such that <sup>∑</sup>*<sup>V</sup>*

Fig. 2. The effects of nonlinear distortion on 16-QAM OFDM constellation for different IBO values

Assuming that the nonlinear distortion is additive and Gaussian, the OFDM signal at the output of the nonlinearity can be written as

$$
\mathfrak{X} = \mathbb{G}\_{\text{HPA}}\mathfrak{x} + d,\tag{3}
$$

where the term *x* is the distortion free input signal vector. *GHPA* is the complex scaling term that is responsible for the attenuation and rotation of the constellation. The term *d* is responsible for clouding of the constellation is the function of the modulated symbol vector *x* and the nonlinear transfer function *g*(·). Moreover, if the symbol size is large and so the number of nonzero distortion terms, the distortion term will be approximately Gaussian random variable, as was already pointed out in Tellado (2000).

The constellation of an exemplary distorted 16-QAM symbol alphabet for selected IBO values is shown in Figure 2. Figure 2 also confirms that in-band nonlinear distortion behaves as an additive Gaussian noise.

#### **4. Review of selected PAPR reduction methods**

In this section, we provide the brief overview of the most well-known PAPR reduction methods. Formerly, they were designed for conventional SISO systems, but the extension to MIMO systems is in the most cases straightforward.

#### **4.1 Clipping of the transmitted signal**

The simplest technique for reduction of the envelope fluctuation is clipping. In clipping all the samples exceeding a given threshold are forced to this maximum value. This is similar to the passing signal through soft limiter nonlinearity. The major disadvantage of clipping technique is that it introduces distortion and increase both BER and out-of-band radiation. In order to improve BER performance, the receiver needs to estimate clipping that has occurred and conversely, compensate received OFDM signal accordingly.

Authors in Kwon et al. (2009) propose the new low complexity SFBC transmitter for clipped OFDM signals, which preserves orthogonality of transmitted signals. Furthermore, clipping reconstruction method for SFBC/STBC-OFDM system based on iterative amplitude reconstruction (IAR) in Kwon & Im (2006) is presented. Another approach for improving the performance of clipped MIMO-OFDM systems with (quasy)-OSTBC is to use the statistics of the clipping distortions to develop maximum likelihood (ML) decoding Li & Xia (2008). For the case of spatially multiplexed systems, the soft correction method of Bittner et al. (2008) is applicable.

#### **4.2 The selected mapping scheme**

4 Will-be-set-by-IN-TECH

IBO = 6 dB

−2.5 −2 −1.5 −1 −0.5 <sup>0</sup> 0.5 <sup>1</sup> 1.5 <sup>2</sup> 2.5 −2.5

Fig. 2. The effects of nonlinear distortion on 16-QAM OFDM constellation for different IBO

Assuming that the nonlinear distortion is additive and Gaussian, the OFDM signal at the

where the term *x* is the distortion free input signal vector. *GHPA* is the complex scaling term that is responsible for the attenuation and rotation of the constellation. The term *d* is responsible for clouding of the constellation is the function of the modulated symbol vector *x* and the nonlinear transfer function *g*(·). Moreover, if the symbol size is large and so the number of nonzero distortion terms, the distortion term will be approximately Gaussian

The constellation of an exemplary distorted 16-QAM symbol alphabet for selected IBO values is shown in Figure 2. Figure 2 also confirms that in-band nonlinear distortion behaves as an

In this section, we provide the brief overview of the most well-known PAPR reduction methods. Formerly, they were designed for conventional SISO systems, but the extension

The simplest technique for reduction of the envelope fluctuation is clipping. In clipping all the samples exceeding a given threshold are forced to this maximum value. This is similar to the passing signal through soft limiter nonlinearity. The major disadvantage of clipping technique is that it introduces distortion and increase both BER and out-of-band radiation. In order to improve BER performance, the receiver needs to estimate clipping that has occurred

Authors in Kwon et al. (2009) propose the new low complexity SFBC transmitter for clipped OFDM signals, which preserves orthogonality of transmitted signals. Furthermore, clipping reconstruction method for SFBC/STBC-OFDM system based on iterative amplitude

*x* = *GHPAx* + *d*, (3)

−2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5

random variable, as was already pointed out in Tellado (2000).

**4. Review of selected PAPR reduction methods**

to MIMO systems is in the most cases straightforward.

and conversely, compensate received OFDM signal accordingly.

**4.1 Clipping of the transmitted signal**

output of the nonlinearity can be written as

additive Gaussian noise.

values

In the selected mapping technique, the transmitter generates several candidate data blocks from the original data block. Subsequently, the one with the lowest envelope fluctuation is transmitted. The candidate data blocks are generated as follows. First, *U* different phase sequences of length *N* are generated, **b**(*u*) = (*b* (*u*) <sup>0</sup> , *b* (*u*) <sup>1</sup> ,..., *b* (*u*) *<sup>N</sup>*−1), *<sup>u</sup>* <sup>=</sup> 1, . . . , *<sup>U</sup>*, where normally **b**<sup>1</sup> is set to be all-one vector of the length *N* in order to include also unmodified block into the set of candidate data blocks. Then candidate data blocks are generated by element-wise multiplication of the frequency-domain OFDM symbol by each **b**(*u*), IFFT is applied and the resulting block with the lowest envelope fluctuations is selected for the transmission. The information about the selected phase sequence has to be transmitted to the receiver in the form of the side information. At the receiver, the reverse operation is performed to recover the original data block.

The straight-forward implementation of conventional SISO SLM is similar to PTS. In this scheme the SISO SLM technique is applied separatelly to each of the *Nt* transmitting antennas in the MIMO-OFDM system. For each of the parallel OFDM frames, the best phase modification out of the *U* possible ones is individually selected.

*Concurrent* SLM approach ensures higher reliability of side information. This is achieved through the spatial diversity by transmitting the same side information on different antennas Lee et al. (2003).

In Fischer & Hoch (2006), authors introduce *directed* SLM (*d*SLM) scheme that uses advantage of multiple antennas. The PAPR decreasing abilities of this method improve with increasing number of antennas, however this comes at the expense of higher number of side information (SI) bits.

In order to improve bandwidth efficiency number of transmitted SI bits has to be decreased. Therefore *small-overhead* SLM was proposed in Hassan et al. (2009). This scheme not only improve bandwidth efficiency but achieves also substantially better BER performance compared to *d*SLM or SISO SLM applied on multiple antennas.

The computational complexity of SLM method is relatively high therefore there is strong need for low-complexity solutions. The promising approach that require only one FFT operation was introduced in Wang & Li (2009). It exploits the time-domain signal properties of MIMO-OFDM systems to achieve a low-complexity architecture for candidate signal generation.

#### **4.3 The partial transmit sequence technique**

In the partial transmit sequence technique Han & Lee (2006); Muller & Huber (1997), the original data block of length *N* is partitioned into *V* disjoint subblocks, **s***<sup>v</sup>* = (*sv*,0,*sv*,1,...,*sv*,*N*−1), *<sup>v</sup>* <sup>=</sup> 1, . . . , *<sup>V</sup>* such that <sup>∑</sup>*<sup>V</sup> <sup>v</sup>*=<sup>1</sup> **s***<sup>v</sup>* = **s**. The subcarriers in each subblock are weighted by a phase factor, *bv* = *ej*Φ*<sup>v</sup>* , *v* = 1, 2, . . . , *V*, for *v*th subblock. Such phase factors are

Recent work in this area includes extensions of the concept of ACE using a modified smart gradient-project (SGP) algorithm for MIMO-OFDM systems Krongold et al. (2005) and extension of the efficient ACE-SGP method to STBC, SFBC and V-BLAST OFDM

Reduction of Nonlinear Distortion in Multi-Antenna WiMAX Systems 65

The basic idea of the tone reservation is to add data-block dependent time domain signal *un* to the original OFDM signal *sn* with aim to reduce its peaks. In case of Tone Reservation (TR), the transmitter does not use the small subset of subcarriers that are reserved for the correcting tones. These reserved subcarriers are then stripped off at the receiver. TR similarly to the ACE

The extension of TR for MIMO-OFDM is straightforward, but it does not take advantage of MIMO potential. TR technique tailored for eigenbeamformed multiple antenna systems has been proposed in Zhang & Goeckel (2007), where authors introduced so-called mode reservation as analogy for TR of SISO-OFDM. Nevertheless this technique requires a perfect

Now, let us assume SFBC-OFDM system with the code rate *r* = 3/4 corresponding to the selected code **C**<sup>334</sup> Jafarkhani (2005) , equipped with *Nt* = 3 transmitting and *Nr* receiving antennas. Furthermore we assume that system employs *Nc* sub-carriers and M-QAM based-band modulation. The data symbol vector **<sup>s</sup>** = [*s*0,*s*1,...,*sr*·*Nc*−1] is encoded with the

<sup>2</sup>, 0, . . . ,*srNc*−3, −*s*

0, 0,*s*2,...,*srNc*−2,*s*

∗ *rNc*−2,*<sup>s</sup>* ∗

> ∗ *rNc*−3,*<sup>s</sup>* ∗

∗

<sup>1</sup>,...,*srNc*−1, 0, −*s*

The vectors **x**1, **x**2, **x**<sup>3</sup> corresponds to the columns of (1). After the mapping according to the orthogonal design on several streams associated with the transmit antennas, a simple serial to parallel converter is used for each transmit antenna, followed by IFFT processing, cyclic prefix

From the above discussion it is clear that due to the SFBC coding scheme, there will be uniformly distributed zero tones at the input of IFFTs. Let us define the positions of the correcting signals Q*R*,*<sup>n</sup>* for *n* = 1, . . . , *Nt* by these zero subcarriers. The proposed method consists of adding the correcting tones at the subcarrier indices occupied by the zero symbols according to Q*R*,*n*, instead of reserving the set of the subcarriers from the data bearing tones. By doing so, we can avoid an important drawback of the tone reservation

It is clear that adding correcting signal to the SFBC encoded signals **x**1, **x**2, **x**<sup>3</sup> may result in loss of the orthogonality, thereby eventually increasing the probability of erroneous detection. The correcting signal represents additive distortion for the decision variables in the receiver. Conversely, in order not to increase BER, the amplitude of the correcting tones must be

*rNc*−1, 0] (5)

*rNc*−2] (7)

*rNc*−3, 0,*srNc*−1] (6)

technique has the slight drawback of the increase in the power of the transmission.

**5. Tone reservation for OFDM SFBC using null subcarriers**

space-frequency encoder producing three vectors **x**1, **x**2, **x**<sup>3</sup> as

∗ <sup>1</sup>,*s* ∗

> ∗ 0, −*s* ∗

insertion and amplification. A simplified block diagram is shown in Figure 3.

∗

**x**<sup>1</sup> = [*s*0, −*s*

**x**<sup>3</sup> = [*s*2, 0, −*s*

**x**<sup>2</sup> = [*s*1,*s*

technique-bandwidth expansion.

systems Tsiligkaridis & Jones (2010).

**4.5 Tone reservation**

CSI at the transmitter.

selected in the way that the envelope fluctuation of the combined signal is minimized. The time domain signal after applying PTS can be expressed as

$$s = \sum\_{v=1}^{V} b\_v \cdot s\_{v\prime} \tag{4}$$

where {*b*1, *b*2, ..., *bV*} is the selected set of phase factors.

The *straight-forward* implementation of the PTS technique for MIMO-OFDM is the independent application of PTS to each transmit antenna. It is just simple application of single antenna PTS. *Simplified approach* provides advantage over straight-forward implementation by decreasing required side information. The input data symbols are converted into the several parallel streams and the conventional PTS technique for single antenna OFDM is applied for each antenna with the sets of the phase factors being equal for all transmit antennas. Since the side information is the same for all transmit antennas, the amount of the side information per transmit antenna is reduced.

Another, *directed PTS*, approach is based on directed Selected Mapping technique Fischer & Hoch (2006). The idea of this technique is to increase number of possible alternative signal representations. In order to keep the complexity similar to the straight-forward or the simplified approach, not all possible candidates are evaluated for each transmitt antenna. The algorithm concentrates on the antenna exhibiting the highest PAPR and aims to reduce it Siegel & Fischer (2008).

In contrast to afore mentioned approaches, *Spatial shifting* provides additional way to exploit presence of multiple antenna by cyclically shifting the partial sequences between antennas Schenk et al. (2005). In other words, instead of using weighting factors for generating the different signal representations cyclic shifting of the partial sequences between the antennas is used. The advantage of this technique is its possible implementation as transparent version, where no side information needs to be transmitted.

Recently, *Siegel* and *Fischer* proposed *Spatially permuted PTS* that is more general permutation compared to cyclic shifting, described above Siegel & Fischer (2008).

Similarly as for SLM method, complexity remains significant issue also for PTS methods. The approach with reduced complexity, named *Polyphase interleaving and inversion*, for SFBC MIMO-OFDM can be find in Latinovi´c & Bar-Ness (2006).

Finally, interesting comparison of PTS and SLM can be found in Siegel & Fischer (2008).The comparison is based on equal computational complexity of both schemes and presented analysis indicate better performance (in terms of PAPR reduction) of PTS method.

#### **4.4 Active constellation extension**

In the active constellation extension strategy, some of the outer constellation points of each OFDM block are extended toward the outside of the original constellation such that the envelope variations of OFDM signal are reduced. By doing this, some constellation points are set to be further from the decision boundaries than the nominal constellation points that slightly reduce BER.

The advantages of ACE are that it is transparent to the receiver, there is no loss of the data rate and no need for the side information. On the other hand, it increases the total transmitted power, that has to be considered in system design.

Recent work in this area includes extensions of the concept of ACE using a modified smart gradient-project (SGP) algorithm for MIMO-OFDM systems Krongold et al. (2005) and extension of the efficient ACE-SGP method to STBC, SFBC and V-BLAST OFDM systems Tsiligkaridis & Jones (2010).

#### **4.5 Tone reservation**

6 Will-be-set-by-IN-TECH

selected in the way that the envelope fluctuation of the combined signal is minimized. The

The *straight-forward* implementation of the PTS technique for MIMO-OFDM is the independent application of PTS to each transmit antenna. It is just simple application of single antenna PTS. *Simplified approach* provides advantage over straight-forward implementation by decreasing required side information. The input data symbols are converted into the several parallel streams and the conventional PTS technique for single antenna OFDM is applied for each antenna with the sets of the phase factors being equal for all transmit antennas. Since the side information is the same for all transmit antennas, the amount of the side information per

Another, *directed PTS*, approach is based on directed Selected Mapping technique Fischer & Hoch (2006). The idea of this technique is to increase number of possible alternative signal representations. In order to keep the complexity similar to the straight-forward or the simplified approach, not all possible candidates are evaluated for each transmitt antenna. The algorithm concentrates on the antenna exhibiting the highest PAPR and aims to reduce it

In contrast to afore mentioned approaches, *Spatial shifting* provides additional way to exploit presence of multiple antenna by cyclically shifting the partial sequences between antennas Schenk et al. (2005). In other words, instead of using weighting factors for generating the different signal representations cyclic shifting of the partial sequences between the antennas is used. The advantage of this technique is its possible implementation as

Recently, *Siegel* and *Fischer* proposed *Spatially permuted PTS* that is more general permutation

Similarly as for SLM method, complexity remains significant issue also for PTS methods. The approach with reduced complexity, named *Polyphase interleaving and inversion*, for SFBC

Finally, interesting comparison of PTS and SLM can be found in Siegel & Fischer (2008).The comparison is based on equal computational complexity of both schemes and presented

In the active constellation extension strategy, some of the outer constellation points of each OFDM block are extended toward the outside of the original constellation such that the envelope variations of OFDM signal are reduced. By doing this, some constellation points are set to be further from the decision boundaries than the nominal constellation points that

The advantages of ACE are that it is transparent to the receiver, there is no loss of the data rate and no need for the side information. On the other hand, it increases the total transmitted

analysis indicate better performance (in terms of PAPR reduction) of PTS method.

transparent version, where no side information needs to be transmitted.

compared to cyclic shifting, described above Siegel & Fischer (2008).

MIMO-OFDM can be find in Latinovi´c & Bar-Ness (2006).

power, that has to be considered in system design.

**4.4 Active constellation extension**

slightly reduce BER.

*bv* · *sv*, (4)

*s* = *V* ∑ *v*=1

time domain signal after applying PTS can be expressed as

where {*b*1, *b*2, ..., *bV*} is the selected set of phase factors.

transmit antenna is reduced.

Siegel & Fischer (2008).

The basic idea of the tone reservation is to add data-block dependent time domain signal *un* to the original OFDM signal *sn* with aim to reduce its peaks. In case of Tone Reservation (TR), the transmitter does not use the small subset of subcarriers that are reserved for the correcting tones. These reserved subcarriers are then stripped off at the receiver. TR similarly to the ACE technique has the slight drawback of the increase in the power of the transmission.

The extension of TR for MIMO-OFDM is straightforward, but it does not take advantage of MIMO potential. TR technique tailored for eigenbeamformed multiple antenna systems has been proposed in Zhang & Goeckel (2007), where authors introduced so-called mode reservation as analogy for TR of SISO-OFDM. Nevertheless this technique requires a perfect CSI at the transmitter.

#### **5. Tone reservation for OFDM SFBC using null subcarriers**

Now, let us assume SFBC-OFDM system with the code rate *r* = 3/4 corresponding to the selected code **C**<sup>334</sup> Jafarkhani (2005) , equipped with *Nt* = 3 transmitting and *Nr* receiving antennas. Furthermore we assume that system employs *Nc* sub-carriers and M-QAM based-band modulation. The data symbol vector **<sup>s</sup>** = [*s*0,*s*1,...,*sr*·*Nc*−1] is encoded with the space-frequency encoder producing three vectors **x**1, **x**2, **x**<sup>3</sup> as

$$\mathbf{x}\_1 = [\mathbf{s}\_{0\prime} - \mathbf{s}\_{1\prime}^\* \mathbf{s}\_{2\prime}^\* \mathbf{0}, \dots, \mathbf{s}\_{rN\_\ell - 3\prime} - \mathbf{s}\_{rN\_\ell - 2\prime}^\* \mathbf{s}\_{rN\_\ell - 1\prime}^\* \mathbf{0}] \tag{5}$$

$$\mathbf{x}\_2 = [\mathbf{s}\_{1\prime}\mathbf{s}\_0^\*, \mathbf{0}, \mathbf{s}\_{2\prime}, \dots, \mathbf{s}\_{rN\_c - 2\prime}\mathbf{s}\_{rN\_c - 3\prime}^\*, \mathbf{0}, \mathbf{s}\_{rN\_c - 1}] \tag{6}$$

$$\mathbf{x}\_3 = \begin{bmatrix} s\_{2\prime}0\_{\prime} - s\_{0\prime}^\* - s\_{1\prime}^\* \dots \imath\_{rN\_{\ell}-1} 0\_{\prime} - s\_{rN\_{\ell}-3\prime}^\* s\_{rN\_{\ell}-2}^\* \end{bmatrix} \tag{7}$$

The vectors **x**1, **x**2, **x**<sup>3</sup> corresponds to the columns of (1). After the mapping according to the orthogonal design on several streams associated with the transmit antennas, a simple serial to parallel converter is used for each transmit antenna, followed by IFFT processing, cyclic prefix insertion and amplification. A simplified block diagram is shown in Figure 3.

From the above discussion it is clear that due to the SFBC coding scheme, there will be uniformly distributed zero tones at the input of IFFTs. Let us define the positions of the correcting signals Q*R*,*<sup>n</sup>* for *n* = 1, . . . , *Nt* by these zero subcarriers. The proposed method consists of adding the correcting tones at the subcarrier indices occupied by the zero symbols according to Q*R*,*n*, instead of reserving the set of the subcarriers from the data bearing tones. By doing so, we can avoid an important drawback of the tone reservation technique-bandwidth expansion.

It is clear that adding correcting signal to the SFBC encoded signals **x**1, **x**2, **x**<sup>3</sup> may result in loss of the orthogonality, thereby eventually increasing the probability of erroneous detection. The correcting signal represents additive distortion for the decision variables in the receiver. Conversely, in order not to increase BER, the amplitude of the correcting tones must be

1 1.5 2 2.5 3 3.5 4 4.5 5

Reduction of Nonlinear Distortion in Multi-Antenna WiMAX Systems 67

QPSK 16 QAM 16 QAM, cc 64 QAM 64 QAM, cc

IBO [dB]

conventional system without the application of TR. Moreover, such a value is suitable for most of the system setup implementations. It can be seen in Figure 5 that the spectrum at the center of the adjacent channel is reduced by 2.7 dB and 4.3 dB when the nonlinearity is operating at IBO = 4dB and 5dB respectively. Based on the analytical results introduced in Deumal et al. (2008) it can be stated that the amount of the out-of-band radiation is independent on the mapping scheme. Therefore by applying the proposed technique here, the same out-of-band radiation suppression can be observed for all modulation formats which make the application

This novel method aims to improve the system performance of SFBC OFDM based transmission system affected by the nonlinear amplification by means of the iterative decoding. It will be showed that the BER performance could be significantly improved even after the first iteration of the decoding process and thus, does not require the large computation processing. Moreover, also the second and the third iteration might be beneficial,

Now, we would like to express the input signal of the receiver in the frequency domain. Let **Y** be the *Nc* × *Nr* matrix containing received signal after CP removal and OFDM demodulation. Similarly to the transmitter case, we can divide **Y** into *Ng* sub-blocks yielding

. Then, the SFBC-OFDM system follows input-output relationship

**Y***<sup>g</sup>* = **X***g***H***<sup>g</sup>* + **W***g*, (8)

Fig. 4. Maximal normalized amplitude of reserved tones for various IBO satisfying

0

of the proposed technique robust in general.

especially in the strong nonlinear propagation environment.

**6. Iterative nonlinear detection**

**Y**0, **Y**1,..., **Y***Ng*−<sup>1</sup>

*BERTR* ≤ *BERconv*

**Y** =  0.1

0.2

0.3

γ

0.4

0.5

0.6

Fig. 3. Transmitter of MIMO SFBC-OFDM employing *C*<sup>334</sup> code

controlled. Maximal amplitude that does not result in increase in BER depends on both, the baseband modulation scheme and the in-band nonlinear distortion introduced by HPA. Figure 4 shows maximum allowed amplitude vs. IBO for various modulations. All curves fulfill the following condition:*BERTR* ≤ *BERconv* i.e. BER of TR based SFBC-OFDM system is lower or equal to that of the conventional system. This figure can be used by system designer as upper bound for the amplitude of the reserved tones in the different system setups. As it can be appreciated, these results are in compliance with our previous assumptions. We can go for higher amplitudes of peak-reduction tones and achieve large out-of-band radiation reduction without BER penalty when QPSK and 16 QAM or coded 64 QAM are adopted for the transmission. The presumptions of the amplitude constraints when uncoded 64 QAM is used are of more relevance, especially for lower IBO. In other words, when applying the uncoded higher modulation schemes (e.g. 64 QAM), the amplitude of the correcting tones is constrained to the very low power, leading to poorer performance of the proposed method performing at the low IBO. However, it should be noted that for low IBO achieved BER of the original system is very poor, characterized by the occurrence of the error floor, thus this performance is not of our interest. Because of this, designer must go for the higher IBO.

Figure 5 shows the PSD of original and TR-reduced OFDM signals when a soft limiter operating at IBOs of 4dB or 5dB is present at the output of the transmitter. In order to prevent the BER performance degradation resulting from the broken space orthogonality among transmitted signals, the maximum amplitude *γ* is constrained to be *γ* = 0.2. That corresponds to the power of reserved tones being more than 14 dB lower than the average signal power. It allows for obtaining the reduction in terms of the out-band-radiation while keeping the BER performance of the system at the same or even better level than BER of the 8 Will-be-set-by-IN-TECH

0

s3


0


controlled. Maximal amplitude that does not result in increase in BER depends on both, the baseband modulation scheme and the in-band nonlinear distortion introduced by HPA. Figure 4 shows maximum allowed amplitude vs. IBO for various modulations. All curves fulfill the following condition:*BERTR* ≤ *BERconv* i.e. BER of TR based SFBC-OFDM system is lower or equal to that of the conventional system. This figure can be used by system designer as upper bound for the amplitude of the reserved tones in the different system setups. As it can be appreciated, these results are in compliance with our previous assumptions. We can go for higher amplitudes of peak-reduction tones and achieve large out-of-band radiation reduction without BER penalty when QPSK and 16 QAM or coded 64 QAM are adopted for the transmission. The presumptions of the amplitude constraints when uncoded 64 QAM is used are of more relevance, especially for lower IBO. In other words, when applying the uncoded higher modulation schemes (e.g. 64 QAM), the amplitude of the correcting tones is constrained to the very low power, leading to poorer performance of the proposed method performing at the low IBO. However, it should be noted that for low IBO achieved BER of the original system is very poor, characterized by the occurrence of the error floor, thus this performance is not of our interest. Because of this, designer must go for the higher IBO. Figure 5 shows the PSD of original and TR-reduced OFDM signals when a soft limiter operating at IBOs of 4dB or 5dB is present at the output of the transmitter. In order to prevent the BER performance degradation resulting from the broken space orthogonality among transmitted signals, the maximum amplitude *γ* is constrained to be *γ* = 0.2. That corresponds to the power of reserved tones being more than 14 dB lower than the average signal power. It allows for obtaining the reduction in terms of the out-band-radiation while keeping the BER performance of the system at the same or even better level than BER of the

s3

s2

s\* 3

0

S/P IFFT P/S

S/P IFFT P/S

S/P IFFT P/S

CP HPA

CP HPA

CP HPA

s1


> s1 \*

Othogonal design

0 s -s s 3 21 s 0s s 3 12 -s -s 0 s 21 3

\* \* \* \* \*

Fig. 3. Transmitter of MIMO SFBC-OFDM employing *C*<sup>334</sup> code

sss <sup>321</sup>

Fig. 4. Maximal normalized amplitude of reserved tones for various IBO satisfying *BERTR* ≤ *BERconv*

conventional system without the application of TR. Moreover, such a value is suitable for most of the system setup implementations. It can be seen in Figure 5 that the spectrum at the center of the adjacent channel is reduced by 2.7 dB and 4.3 dB when the nonlinearity is operating at IBO = 4dB and 5dB respectively. Based on the analytical results introduced in Deumal et al. (2008) it can be stated that the amount of the out-of-band radiation is independent on the mapping scheme. Therefore by applying the proposed technique here, the same out-of-band radiation suppression can be observed for all modulation formats which make the application of the proposed technique robust in general.

#### **6. Iterative nonlinear detection**

This novel method aims to improve the system performance of SFBC OFDM based transmission system affected by the nonlinear amplification by means of the iterative decoding. It will be showed that the BER performance could be significantly improved even after the first iteration of the decoding process and thus, does not require the large computation processing. Moreover, also the second and the third iteration might be beneficial, especially in the strong nonlinear propagation environment.

Now, we would like to express the input signal of the receiver in the frequency domain. Let **Y** be the *Nc* × *Nr* matrix containing received signal after CP removal and OFDM demodulation. Similarly to the transmitter case, we can divide **Y** into *Ng* sub-blocks yielding **Y** = **Y**0, **Y**1,..., **Y***Ng*−<sup>1</sup> . Then, the SFBC-OFDM system follows input-output relationship

$$\mathbf{Y}\_{\mathcal{S}} = \mathbf{X}\_{\mathcal{S}} \mathbf{H}\_{\mathcal{S}} + \mathbf{W}\_{\mathcal{S}'} \tag{8}$$

OFDM-1 SFBC

can be estimated from the received symbol vector **Y***g*.

(*i*)

to signals at the output of SFBC decoder according :

respectively. The estimated distortion terms ˜*d*

*X <sup>g</sup>* = *g X*˜ *g* .

procedure will consist of the following steps:

distorted signals

receiver does not know **D**(**H**)

1. Compute the estimation *s*˜

encoding and ˜

view.

3. Go to step 1 and compute *s*˜

Term **D**�

combining

OFDM

*<sup>g</sup>* . However, if receiver knows the transmit nonlinear function, it

*<sup>g</sup>*,*<sup>k</sup>* of the transmitted symbol *sg*,*<sup>k</sup>* by the hard decisions applied

*<sup>g</sup>*,*<sup>k</sup>* are assumed to be zero for *i* = 1.

 *s*˜ (*i*) *<sup>g</sup>*,0,...,*s*˜

(*i*) *g*,*K*−1 

HPA model

Let us assume, that complex characteristics of HPA *g*(·) and channel frequency responses are known. Then, taking into account these assumptions, the nonlinear iterative detection

*<sup>y</sup>*˜*g*,*<sup>k</sup>* <sup>−</sup> ˜*<sup>d</sup>*

The symbols *<* · *>* and *i* denote the hard decision operation and the iteration number,

**<sup>D</sup>**˜ *<sup>g</sup>* <sup>=</sup> *FFT* ˜

The block scheme of the proposed iterative receiver is depicted in Fig. 6. The iterative process is stopped if *BER*(*i* + 1) = *BER*(*i*) or if the BER is acceptable from an application point of

Figure 7 shows the performance of the proposed method for different iterations with {16, 64}-QAM and Rapp model of HPA operating at IBO = 5 dB. We assume convolutionaly coded system. Most of the performance improvement is achieved with first and second

(*i*)

(*i*−1) *g*,*k* 

*<sup>X</sup> <sup>g</sup>* <sup>−</sup> *<sup>X</sup>*˜ *<sup>g</sup>*

(*i*) *<sup>g</sup>* =


OFDM-1

Distortion

Reduction of Nonlinear Distortion in Multi-Antenna WiMAX Systems 69


Calculation CSI

Fig. 6. Proposed SFBC-OFDM receiver structure for iterative detection of nonlinearly

*<sup>g</sup>* is obtained from **<sup>D</sup>***<sup>g</sup>* by conjugating second half of **<sup>D</sup>**(**H**)

*s*˜ (*i*) *<sup>g</sup>*,*<sup>k</sup>* = 

2. Compute the estimation **D**˜ *<sup>g</sup>* of the nonlinear distortion terms **D***<sup>g</sup>*

where *X*˜ *<sup>g</sup>* is obtained by taking the IFFT of block **s˜**

(*i*+1) *<sup>g</sup>*,*<sup>k</sup>* .

SFBC encoding

Hard Decision Demod.

*<sup>g</sup>* entries. In practice the

(13)

after SFBC

Fig. 5. PSD of a conventional and a TR-based SFBC-OFDM system obtained when a soft limiter is present. IBO={4, 5} dB.

for *g* = 0, 1, . . . , *Ng* − 1. The **W***<sup>g</sup>* is *Ns* × *Nr* matrix containing noise samples with variance *σ*2 *<sup>n</sup>* and **H***<sup>g</sup>* is *Nt* × *Nr* matrix of path gains *hn* between *n* − *th* transmit and receive antenna at subcarrier frequency *g* · *Ns*.

From (3) and (8), the signal in the frequency domain at the output of OFDM demodulator can be rewritten as

$$\mathbf{Y}\_{\mathcal{S}} = (\mathbf{X}\_{\mathcal{S}} + \mathbf{D}\_{\mathcal{S}})\mathbf{H}\_{\mathcal{S}} + \mathbf{W}\_{\mathcal{S}'} \tag{9}$$

where noise term **D***<sup>g</sup>* is the frequency domain representation of nonlinear distortion. Hence, the maximum likelihood sequence detector has to find codeword **X**˜ *<sup>g</sup>* that minimises frobenius norm as

$$\tilde{\mathbf{X}}\_{\mathcal{S}} = \arg\min\_{\forall \tilde{\mathbf{X}}\_{\mathcal{S}}} \left\| \left| \mathbf{Y}\_{\mathcal{S}} - \left( \check{\mathbf{X}}\_{\mathcal{S}} \mathbf{H}\_{\mathcal{S}} + \mathbf{D}\_{\mathcal{S}} \mathbf{H}\_{\mathcal{S}} \right) \right| \right\|\_{F'} \tag{10}$$

where **X**ˇ *<sup>g</sup>* is any possible transmitted codeword Drotár et al. (2010b). Using a full search to find the optimal codeword is computationally very demanding. However, if we assume that receiver knows NLD it can be compensated in decision variables. Since **D***<sup>g</sup>* is deterministic it does not play any role in ML detector. Orthogonal SFBC coding structure that we have considered make it possible to implement a simpler per-symbol ML decoding Giannakis et al. (2007); Tarokh et al. (1999). It can be shown Drotár et al. (2010b) that transmitted symbols to be decoded separately with small computional complexity as follows

$$\tilde{s}\_{\mathcal{g},k} = \arg\min\_{\forall \mathcal{S}} \left| \left| \tilde{y}\_{\mathcal{g},k} - d\_{\mathcal{g},k} - \kappa \sum\_{n=1}^{N\_l} |h\_n|^2 \check{s}\_{\mathcal{g},k} \right| \right|. \tag{11}$$

Here, *<sup>y</sup>*˜*g*,*<sup>k</sup>* is *<sup>k</sup>* <sup>−</sup> *th* entry of **Y˜** *<sup>g</sup>* and *dg*,*<sup>k</sup>* is *<sup>k</sup>* <sup>−</sup> *th* entry of **<sup>d</sup>***<sup>g</sup>* computed as

$$\mathbf{d}\_{\mathcal{S}} = \mathbf{D}\_{\mathcal{S}}^{'} \mathbf{H}\_{\mathcal{S}}^{H}. \tag{12}$$

10 Will-be-set-by-IN-TECH

Conventional SFBC−OFDM TR based SFBC−OFDM

IBO = 4dB

IBO=5dB

**Y***<sup>g</sup>* = (**X***<sup>g</sup>* + **D***g*)**H***<sup>g</sup>* + **W***g*, (9)

 *F*

> 

*<sup>g</sup>* . (12)

, (10)

. (11)

−2 −1.5 −1 −0.5 0 0.5 1 1.5 2

for *g* = 0, 1, . . . , *Ng* − 1. The **W***<sup>g</sup>* is *Ns* × *Nr* matrix containing noise samples with variance

*<sup>n</sup>* and **H***<sup>g</sup>* is *Nt* × *Nr* matrix of path gains *hn* between *n* − *th* transmit and receive antenna at

From (3) and (8), the signal in the frequency domain at the output of OFDM demodulator can

where noise term **D***<sup>g</sup>* is the frequency domain representation of nonlinear distortion. Hence, the maximum likelihood sequence detector has to find codeword **X**˜ *<sup>g</sup>* that minimises frobenius

where **X**ˇ *<sup>g</sup>* is any possible transmitted codeword Drotár et al. (2010b). Using a full search to find the optimal codeword is computationally very demanding. However, if we assume that receiver knows NLD it can be compensated in decision variables. Since **D***<sup>g</sup>* is deterministic it does not play any role in ML detector. Orthogonal SFBC coding structure that we have considered make it possible to implement a simpler per-symbol ML decoding Giannakis et al. (2007); Tarokh et al. (1999). It can be shown Drotár et al. (2010b) that transmitted symbols to

*y*˜*g*,*<sup>k</sup>* − *dg*,*<sup>k</sup>* − *κ*

*g***H***<sup>H</sup>*

**d***<sup>g</sup>* = **D**�

**Xˇ** *<sup>g</sup>***H***<sup>g</sup>* + **D***g***H***<sup>g</sup>*

*Nt* ∑ *n*=1 |*hn*| <sup>2</sup> *<sup>s</sup>*ˇ*g*,*<sup>k</sup>* 

**X˜** *<sup>g</sup>* = arg min

<sup>∀</sup>**Xˇ** *<sup>g</sup>* **Y***<sup>g</sup>* − 

be decoded separately with small computional complexity as follows

Here, *<sup>y</sup>*˜*g*,*<sup>k</sup>* is *<sup>k</sup>* <sup>−</sup> *th* entry of **Y˜** *<sup>g</sup>* and *dg*,*<sup>k</sup>* is *<sup>k</sup>* <sup>−</sup> *th* entry of **<sup>d</sup>***<sup>g</sup>* computed as

   

*<sup>s</sup>*˜*g*,*<sup>k</sup>* <sup>=</sup> arg min∀*s*<sup>ˇ</sup>

Fig. 5. PSD of a conventional and a TR-based SFBC-OFDM system obtained when a soft

Frequency (normalized to BW)

−50

limiter is present. IBO={4, 5} dB.

subcarrier frequency *g* · *Ns*.

be rewritten as

norm as

*σ*2

−40

−30

PSD [dB]

−20

−10

0

Fig. 6. Proposed SFBC-OFDM receiver structure for iterative detection of nonlinearly distorted signals

Term **D**� *<sup>g</sup>* is obtained from **<sup>D</sup>***<sup>g</sup>* by conjugating second half of **<sup>D</sup>**(**H**) *<sup>g</sup>* entries. In practice the receiver does not know **D**(**H**) *<sup>g</sup>* . However, if receiver knows the transmit nonlinear function, it can be estimated from the received symbol vector **Y***g*.

Let us assume, that complex characteristics of HPA *g*(·) and channel frequency responses are known. Then, taking into account these assumptions, the nonlinear iterative detection procedure will consist of the following steps:

1. Compute the estimation *s*˜ (*i*) *<sup>g</sup>*,*<sup>k</sup>* of the transmitted symbol *sg*,*<sup>k</sup>* by the hard decisions applied to signals at the output of SFBC decoder according :

$$
\mathfrak{s}\_{\mathcal{g},k}^{(i)} = \left< \mathfrak{y}\_{\mathcal{g},k} - \mathfrak{d}\_{\mathcal{g},k}^{(i-1)} \right> \tag{13}
$$

The symbols *<* · *>* and *i* denote the hard decision operation and the iteration number, respectively. The estimated distortion terms ˜*d* (*i*) *<sup>g</sup>*,*<sup>k</sup>* are assumed to be zero for *i* = 1.

2. Compute the estimation **D**˜ *<sup>g</sup>* of the nonlinear distortion terms **D***<sup>g</sup>*

$$
\tilde{\mathbf{D}}\_{\mathcal{S}} = FFT \left( \tilde{\mathbf{\tilde{X}}}\_{\mathcal{S}} - \tilde{\mathbf{x}}\_{\mathcal{S}} \right),
$$

where *X*˜ *<sup>g</sup>* is obtained by taking the IFFT of block **s˜** (*i*) *<sup>g</sup>* = *s*˜ (*i*) *<sup>g</sup>*,0,...,*s*˜ (*i*) *g*,*K*−1 after SFBC encoding and ˜ *X <sup>g</sup>* = *g X*˜ *g* .

3. Go to step 1 and compute *s*˜ (*i*+1) *<sup>g</sup>*,*<sup>k</sup>* .

The block scheme of the proposed iterative receiver is depicted in Fig. 6. The iterative process is stopped if *BER*(*i* + 1) = *BER*(*i*) or if the BER is acceptable from an application point of view.

Figure 7 shows the performance of the proposed method for different iterations with {16, 64}-QAM and Rapp model of HPA operating at IBO = 5 dB. We assume convolutionaly coded system. Most of the performance improvement is achieved with first and second

OFDM-1 ZF/

SM MIMO-OFDM.

receiver.

detector.

symbol at the output of the detector.

improvement even with the first iteration.

MMSE

OFDM

Spatial Multiplexing

HPA model

Fig. 8. Proposed receiver structure for iterative detection of nonlinearly distorted signals in

2. The estimation of transmitted symbol is computed by means of hard decision applied to

3. Further, transmitter processing is modelled in order to obtain estimate of transmitted symbol that allows to compute distortion term, when HPA characteristics is known at the

4. Finally, distortion term in frequency domain is subtracted from the signal at the output of

To evaluate the performance of the proposed detection, let us consider the coded SM MIMO-OFDM system with *Nc* = 128 subcarriers and 2 transmit and 2 receive antennas performing with Rapp nonlinearity. Figure 9 shows the simulation results for Rapp nonlinearity operating at IBO=4 dB using 16-QAM. The results are reported for 1, 2, 3 iterations of proposed cancellation technique. The results of conventional receiver are also shown as a reference. It can be seen that proposed technique provides a serious performance

**7.2 Application to improve BER of tone reservation for SFBC OFDM using null subcarriers** As was indicated in section 5 addition of correcting signal to the SFBC encoded signals may result in loss of orthogonality, thereby eventually degradate BER performance of the system. The probability of erroneous detection is increased because correcting signal represents additive distortion - tone reservation distortion (TRD). In this section, we attempt to cancel

Let us recall from section 5, the SFBC coded signal vectors **x***n*, for *n* = 1, . . . , *Nt* to be transmitted from *Nt* antennas in parallel at *Nc* subcarriers. These signals carry zero symbols at subcarriers positions defined by Q*R*,*n*. The correcting signal in frequency domain **u***<sup>n</sup>* is added to the data signal. The position of nonzero correcting symbols in **u***<sup>n</sup>* is given by Q*R*,*n*.

**x***<sup>n</sup>* + **u***n*. (14)

5. Whole procedure can be repeated to obtain additional improvement.

this distortion at the receiver side of SFBC-OFDM transmission system.

Therefore, the signal to be transmitted from *n*-th antenna can be described as


OFDM-1


Reduction of Nonlinear Distortion in Multi-Antenna WiMAX Systems 71

Hard Decision Demod.

iteration for 16-QAM and 64-QAM, respectively. When more iterations are applied, no further performance improvement is observed. Incremental gains diminish after the first for 16-QAM and second iteration for 64-QAM, respectively. This can be explained by the reasoning that some OFDM blocks are too badly distorted for the iterative process to converge and more iterations will not help.

Fig. 7. BER performance of a coded SFBC-OFDM system with a Rapp nonlinearity operating at IBO=5 dB for {16, 64}-QAM and for {1, 2, 3 } of iterations. HPA characteristics is perfectly known at the receiver.

#### **7. Extension of iterative nonlinear detection**

#### **7.1 Spatial multiplexing**

In the previous section, we have assumed MIMO SFBC-OFDM systems. However, if our aim is to increase capacity of system better solution is to use Spatial Multiplexing (SM) MIMO-OFDM systems. Unfortunately, as long as the fundamental operation of SM MIMO-OFDM remains identical to conventional OFDM, the SM MIMO-OFDM transmitted signal suffers from nonlinear distortion.

It was shown that we can estimate distortion term by using received signal and characteristic of HPA. The estimated distortion term can be afterwards cancelled from the received distorted signal. When the estimation is quite accurate cancellation results in reduction of in-band nonlinear distortion. The very similar approach can be taken also for SM MIMO-OFDM systems.

The procedure of iterative detection is illustrated in Figure 8 and can be described as follows:

1. First, received signal is processed in OFDM demodulator followed by equalisation technique such as zero forcing or minimum mean square error.

12 Will-be-set-by-IN-TECH

iteration for 16-QAM and 64-QAM, respectively. When more iterations are applied, no further performance improvement is observed. Incremental gains diminish after the first for 16-QAM and second iteration for 64-QAM, respectively. This can be explained by the reasoning that some OFDM blocks are too badly distorted for the iterative process to converge and more

<sup>10</sup> <sup>15</sup> <sup>20</sup> <sup>25</sup> <sup>30</sup> <sup>35</sup> <sup>40</sup> 10−5

Eb /N0 [dB]

Fig. 7. BER performance of a coded SFBC-OFDM system with a Rapp nonlinearity operating at IBO=5 dB for {16, 64}-QAM and for {1, 2, 3 } of iterations. HPA characteristics is perfectly

In the previous section, we have assumed MIMO SFBC-OFDM systems. However, if our aim is to increase capacity of system better solution is to use Spatial Multiplexing (SM) MIMO-OFDM systems. Unfortunately, as long as the fundamental operation of SM MIMO-OFDM remains identical to conventional OFDM, the SM MIMO-OFDM transmitted

It was shown that we can estimate distortion term by using received signal and characteristic of HPA. The estimated distortion term can be afterwards cancelled from the received distorted signal. When the estimation is quite accurate cancellation results in reduction of in-band nonlinear distortion. The very similar approach can be taken also for SM MIMO-OFDM

The procedure of iterative detection is illustrated in Figure 8 and can be described as follows: 1. First, received signal is processed in OFDM demodulator followed by equalisation

technique such as zero forcing or minimum mean square error.

iterations will not help.

10−4

known at the receiver.

**7.1 Spatial multiplexing**

systems.

10−3

16−QAM 64−QAM linear HPA conventional rec.(0 it.) 1st iteration 2nd iteration 3rd iteration

**7. Extension of iterative nonlinear detection**

signal suffers from nonlinear distortion.

BER

10−2

10−1

100

Fig. 8. Proposed receiver structure for iterative detection of nonlinearly distorted signals in SM MIMO-OFDM.


To evaluate the performance of the proposed detection, let us consider the coded SM MIMO-OFDM system with *Nc* = 128 subcarriers and 2 transmit and 2 receive antennas performing with Rapp nonlinearity. Figure 9 shows the simulation results for Rapp nonlinearity operating at IBO=4 dB using 16-QAM. The results are reported for 1, 2, 3 iterations of proposed cancellation technique. The results of conventional receiver are also shown as a reference. It can be seen that proposed technique provides a serious performance improvement even with the first iteration.

#### **7.2 Application to improve BER of tone reservation for SFBC OFDM using null subcarriers**

As was indicated in section 5 addition of correcting signal to the SFBC encoded signals may result in loss of orthogonality, thereby eventually degradate BER performance of the system. The probability of erroneous detection is increased because correcting signal represents additive distortion - tone reservation distortion (TRD). In this section, we attempt to cancel this distortion at the receiver side of SFBC-OFDM transmission system.

Let us recall from section 5, the SFBC coded signal vectors **x***n*, for *n* = 1, . . . , *Nt* to be transmitted from *Nt* antennas in parallel at *Nc* subcarriers. These signals carry zero symbols at subcarriers positions defined by Q*R*,*n*. The correcting signal in frequency domain **u***<sup>n</sup>* is added to the data signal. The position of nonzero correcting symbols in **u***<sup>n</sup>* is given by Q*R*,*n*. Therefore, the signal to be transmitted from *n*-th antenna can be described as

1 1.5 2 2.5 3 3.5 4 4.5 5 5.5

IBO [dB]

Reduction of Nonlinear Distortion in Multi-Antenna WiMAX Systems 73

case, the both techniques for reduction of nonlinear distortion introduced in this thesis i.e. tone reservation with no spectral broadening and the iterative receiver technique are applied. BER curves for assumed scenario are depicted in Figure 11. As reported results indicate the best BER performance is achieved when the iterative receiver for estimation and cancellation

> linear HPA conventional it. NLD canc. TR

TR + it. TRD canc. TR + it. NLD canc.

TR + it. TRD cancel. + it. NLD canc.

10 15 20 25 30 35 40

Eb /N0 [dB]

Fig. 11. BER vs. *Eb*/*N*<sup>0</sup> for uncoded SFBC-OFDM employing three transmit antennas and **C**<sup>334</sup> code. Rapp model of HPA operating at IBO=5. HPA characteristics is perfectly known at

Fig. 10. Maximal normalized amplitude of reserved tones for various IBO satisfying

*BERTR* ≤ *BERconv*, TRD cancellation technique applied at the receiver

0

10<sup>−</sup><sup>5</sup>

the receiver.

10<sup>−</sup><sup>4</sup>

10<sup>−</sup><sup>3</sup>

BER

10<sup>−</sup><sup>2</sup>

10<sup>−</sup><sup>1</sup>

100

0.1

0.2

0.3

γ

0.4

0.5

0.6

−QAM, TRD canc. −QAM, TRD canc. −QAM, cc, TRD canc. −QAM, cc, TRD. canc.

Fig. 9. BER performance of a coded SM MIMO-OFDM system with a Rapp nonlinearity operating at IBO=4 dB, 16-QAM and for {1, 2, 3 } iterations. HPA characteristic is perfectly known at the receiver.

Let us assume only one receive antenna. Then, the received signal in the frequency domain is

$$\mathbf{Y} = \sum\_{n=1}^{N\_l} \left( \mathbf{x}\_{\text{ll}} + \mathbf{u}\_{\text{ll}} + \mathbf{d}\_{\text{ll}} \right) \odot \mathbf{h}\_{\text{ll}} + \mathbf{w}\_{\text{ll}}.\tag{15}$$

Here **d***<sup>n</sup>* represents the in-band nonlinear distortion, **h***<sup>n</sup>* is the channel frequency response between *n*-th transmit and receive antenna, **w** is vector of AWGN noise samples and stands for element-wise multiplication. The best way how to limit the influence of TRD, represented by **u***n*, on decision variable is to cancel it from received signal. However, in order to subtract TRD from received signal correcting signal has to be known. The feasible approach is to obtain the estimate of correcting signal by means of iterative estimation and then cancel it from received signal. The background and details of process of iterative estimation and cancellation were treated in detail in the section 6 for the matter of nonlinear distortion. Now, we will apply the same concept in the straight-forward manner for TRD.

Similarly to Figure 4, in Figure 10 we show the maximal available amplitudes of correcting signal, that can be used in conjunction with TRD cancellation technique. As it can be seen from Figure 10 the combination of TRD cancellation and convolutional coding for 64-QAM leads to higher affordable amplitudes in comparison with only coding application. Moreover, the combination of these approaches makes it possible to use TR technique with no spectral broadening also for 256-QAM modulation.

Finally, we present performance results for uncoded SFBC-OFDM employing three transmit antennas and **C**<sup>334</sup> code. Rapp model of the HPA operating at IBO=5 dB is assumed. In this 14 Will-be-set-by-IN-TECH

linear HPA conventional rx(0 it.) 1st iteration 2nd iteration 3rd iteration

(**x***<sup>n</sup>* + **u***<sup>n</sup>* + **d***n*) **h***<sup>n</sup>* + **w***n*. (15)

10 15 20 25 30 35 40

Eb /N0

Let us assume only one receive antenna. Then, the received signal in the frequency domain is

Here **d***<sup>n</sup>* represents the in-band nonlinear distortion, **h***<sup>n</sup>* is the channel frequency response between *n*-th transmit and receive antenna, **w** is vector of AWGN noise samples and stands for element-wise multiplication. The best way how to limit the influence of TRD, represented by **u***n*, on decision variable is to cancel it from received signal. However, in order to subtract TRD from received signal correcting signal has to be known. The feasible approach is to obtain the estimate of correcting signal by means of iterative estimation and then cancel it from received signal. The background and details of process of iterative estimation and cancellation were treated in detail in the section 6 for the matter of nonlinear distortion. Now, we will apply

Similarly to Figure 4, in Figure 10 we show the maximal available amplitudes of correcting signal, that can be used in conjunction with TRD cancellation technique. As it can be seen from Figure 10 the combination of TRD cancellation and convolutional coding for 64-QAM leads to higher affordable amplitudes in comparison with only coding application. Moreover, the combination of these approaches makes it possible to use TR technique with no spectral

Finally, we present performance results for uncoded SFBC-OFDM employing three transmit antennas and **C**<sup>334</sup> code. Rapp model of the HPA operating at IBO=5 dB is assumed. In this

Fig. 9. BER performance of a coded SM MIMO-OFDM system with a Rapp nonlinearity operating at IBO=4 dB, 16-QAM and for {1, 2, 3 } iterations. HPA characteristic is perfectly

**Y** =

the same concept in the straight-forward manner for TRD.

broadening also for 256-QAM modulation.

*Nt* ∑ *n*=1

10−5

known at the receiver.

10−4

10−3

BER

10−2

10−1

100

Fig. 10. Maximal normalized amplitude of reserved tones for various IBO satisfying *BERTR* ≤ *BERconv*, TRD cancellation technique applied at the receiver

case, the both techniques for reduction of nonlinear distortion introduced in this thesis i.e. tone reservation with no spectral broadening and the iterative receiver technique are applied. BER curves for assumed scenario are depicted in Figure 11. As reported results indicate the best BER performance is achieved when the iterative receiver for estimation and cancellation

Fig. 11. BER vs. *Eb*/*N*<sup>0</sup> for uncoded SFBC-OFDM employing three transmit antennas and **C**<sup>334</sup> code. Rapp model of HPA operating at IBO=5. HPA characteristics is perfectly known at the receiver.

Drotár, P., Gazda, J. & et. al. (2010b). Receiver technique for iterative estimation and

Reduction of Nonlinear Distortion in Multi-Antenna WiMAX Systems 75

Fischer, R. F. & Hoch, M. (2006). Directed selected mapping for peak-to-average power ratio

Giannakis, G. B., Liu, Z., Ma, X. & Zhou, S. (2007). *Space-time coding for broadband wireless*

Han, S. H. & Lee, J. H. (2006). PAPR reduction of OFDM signals using a reduced complexity

Hassan, E., El-Khamy, S., Dessouky, M., El-Dolil, S. & El-Samie, F. A. (2009). Peak-to-average

Jafarkhani, H. (2005). *Space - Time Coding: Theory and Practice*, Cambridge University Press,

Krongold, B. S., Woo, G. R. & et. al. (2005). Fast active constellation extension for

Kwon, U. I., Kim, K. & Im, G.-H. (2009). Amplitude clipping and iterative reconstruction

Kwon, U. K. & Im, G. H. (2006). Iterative amplitude reconstruction of clipped OFDM signals

Latinovi´c, Z. & Bar-Ness, Y. (2006). SFBC MIMO-OFDM peak-to-average power

Lee, Y.-L., You, Y.-H. & et.al. (2003). Peak-to-average power ratio in MIMO-OFDM systems

Li, Z. & Xia, X.-G. (2008). Single-symbol ML decoding for orthogonal and quasi-orthogonal

Liu, Z., Xin, Y. & Giannakis, G. B. (2002). Space-time-frequency coded OFDM over

Muller, S. H. & Huber, J. B. (1997). OFDM with reduced peak-to-average power ratio by optimum combination of partial transmit sequences, *IEE Electronics Lett.* 33: 36–39. Ochiai, H. & Imai, H. (2001). On the distribution of the peak-to-average power ratio in OFDM

Schenk, T., Smulders, P. & Fledderus, E. (2005). Peak-to-average power reduction in space

Siegel, C. & Fischer, R. F. H. (2008). Partial transmit sequences for peak-to-average power ratio

with optimum equalization, *IEE Electronics Lett.* 42: 1189–1190.

using selective mapping, *IEEE Commun. letters* 7: 575–577.

approximation, *IEEE Trans. Commun.* 56: 1127–1137.

signals, *IEEE Trans. Commun.* 49: 282–289.

Khan, F. (2009). *LTE for 4G mobile broadband*, Cambridge University Press, Cambridge.

power ratio reduction in spaceUtime block coded multi-input multi-output ˝ orthogonal frequency division multiplexing systems using a small overhead selective

MIMO-OFDM PAR reduction, *Proc. IEEE Int. Conference on Communications*, pp. 1476

of MIMO-OFDM signals with optimum equalization, *IEEE Trans. Wireless Commun.*

ratio reduction by polyphase interleaving and inversion, *IEEE Commun. letters*

stbc in clipped MIMO-OFDM systems using a clipping noise model with gaussian

frequency-selective fading channels, *IEEE Trans. on Signal Processing* 50: 2465–2476.

division multiplexing based OFDM systems through spatial shifting, *Electronic letters*

reduction in multiantenna OFDM, *EURASIP Journal on Wireless Communications and*

reduction in MIMO OFDM, *IEE Electronics Lett.* 42: 1289–1290.

*communications*, John Wiley & Sons, Hoboken, USA.

PTS technique, *IEEE Signal Process. Lett.* 11: 887–890.

mapping scheme, *IET Communications* 3: 1667–1674.

*Consum. Electron.* 56: 10–16.

New York, USA.

– 1479.

8(1): 268–277.

10: 266–268.

41: 860–861.

*Networking* 2008: 1–11.

cancellation of nonlinear distortion in MIMO SFBC-OFDM systems, *IEEE Trans.*

of NLD (it. NLD canc.) is used. This is illustrated by a curve with circle marker. However, applying only the receiver technique does not bring any reduction in out-of-band radiation at the transmitter side. Therefore, TR with no spectral broadening was applied at the transmitter. Amplitude of correcting tones was constraint to *γ* = 0.2, but this results in increased BER for the Rapp nonlinearity operating at IBO=5 dB. Increase in BER is noticeable for TR with no spectral broadening when compared to the conventional system and also for application of TR together with iterative NLD cancellation compared to iterative NLD cancellation without TR. Fortunately, this can be solved by application of the receiver cancellation of TRD. Then, the dotted marker BER curve represents results for the application of both the transmitter and the receiver based methods. As can be seen from the figure significant BER performance reduction is obtained, moreover out-of-band radiation reduction is also achieved.

#### **8. Conclusion**

This chapter deals with the nonlinear impairments occuring in OFDM MIMO transmission. We present the brief overview of several PAPR reduction methods. The major contribution of this chapter is the introduction of two strategies, capable of mitigating the nonlinear impairments occuring in MIMO OFDM based transmission system. The fundamental idea of the former one is to use the null subcarriers for the reduction of the out-of-band radiation. The latter method, employed in the detector, improves significantly the BER performance of the MIMO-OFDM system degradaded by HPA nonlinearities. Finally, we present their joint impact on overall performance of MIMO-OFDM sytem operating over nonlinear channel. We show that the application of these methods is specially vital in the broadcast cellular standards, such as WiMAX, and therefore we believe that this contribution might be of interest to the readers and researchers working in this area.

#### **9. Acknowledgments**

Work was supported by VEGA Advanced Signal Processing Techniques for Reconfigurable Wireless Sensor Networks, VEGA 1/0045/10, 2010 U 2011. ˝

#### **10. References**


16 Will-be-set-by-IN-TECH

of NLD (it. NLD canc.) is used. This is illustrated by a curve with circle marker. However, applying only the receiver technique does not bring any reduction in out-of-band radiation at the transmitter side. Therefore, TR with no spectral broadening was applied at the transmitter. Amplitude of correcting tones was constraint to *γ* = 0.2, but this results in increased BER for the Rapp nonlinearity operating at IBO=5 dB. Increase in BER is noticeable for TR with no spectral broadening when compared to the conventional system and also for application of TR together with iterative NLD cancellation compared to iterative NLD cancellation without TR. Fortunately, this can be solved by application of the receiver cancellation of TRD. Then, the dotted marker BER curve represents results for the application of both the transmitter and the receiver based methods. As can be seen from the figure significant BER performance

reduction is obtained, moreover out-of-band radiation reduction is also achieved.

to the readers and researchers working in this area.

Beijing, China, pp. 609–614.

47(1): 137–147.

Wireless Sensor Networks, VEGA 1/0045/10, 2010 U 2011.

This chapter deals with the nonlinear impairments occuring in OFDM MIMO transmission. We present the brief overview of several PAPR reduction methods. The major contribution of this chapter is the introduction of two strategies, capable of mitigating the nonlinear impairments occuring in MIMO OFDM based transmission system. The fundamental idea of the former one is to use the null subcarriers for the reduction of the out-of-band radiation. The latter method, employed in the detector, improves significantly the BER performance of the MIMO-OFDM system degradaded by HPA nonlinearities. Finally, we present their joint impact on overall performance of MIMO-OFDM sytem operating over nonlinear channel. We show that the application of these methods is specially vital in the broadcast cellular standards, such as WiMAX, and therefore we believe that this contribution might be of interest

Work was supported by VEGA Advanced Signal Processing Techniques for Reconfigurable

Baytekin, B. & Meyer, R. G. (2005). Analysis and simulation of spectral regrowth in radio

Bittner, S., Zillmann, P. & Fettweis, G. (2008). Equalisation of MIMO-OFDM signals affected

Dardari, D., Tralli, V. & Vaccari., A. (2000). A theoretical characterization of nonlinear distortion effects in OFDM systems, *IEEE Trans. Commun.* 48: 1755–1764. Deumal, M., Behravan, A., Eriksson, T. & Pijoan, J. (2008). Evaluation of performance

Drotar, P., Gazda, J. & et. al. (2010a). Receiver based compensation of nonlinear distortion in

*Defined and Cognitive Radio Solutions*, Aveiro, Portugal, pp. 53–57.

by phase noise and clipping and filtering, *Proc. IEEE Int. Conf on Communications*,

capabilities of PAPR reducing methods, *Wireless Personal Communications*

MIMO-OFDM, *Proc IEEE Int. Microwave Workshop Series on RF Front-ends for Software*

frequency power amplifiers, *IEEE J. Solid-State Circuits* 40: 370–381.

˝

**8. Conclusion**

**9. Acknowledgments**

**10. References**


**5** 

*Romania* 

**MicroTCA Compliant WiMAX BS** 

Modern mobile communication systems must fulfill more and more requirements received from the customers. This leads to an increase of complexity. The control part of the system becomes very important, a multi-level approach being needed. With respect to this, all BS (Base Stations) from a system are synchronized using GPS (Global Positioning System) or IEEE 1588 [1] standard, high speed synchronous interfaces are used between the BBM (Baseband Modules) and the RRU (Remote Radio Units), for example OBSAI (Open Base Station Architecture Initiative) [2, 3] or CPRI (Common Public Radio Interface) [4], and standard communication methods are provided between the control parts placed in

This chapter describes the management and synchronization procedures for a WiMAX BS architecture compliant with MicroTCA standard (Micro Telecommunications Computing Architecture) [5]. The block scheme of such a BS for the case of a 3 sectors cell is presented. One can observe the main parts of the MicroTCA standard, i.e. the MCH (MicroTCA Carrier

Referring now to the OBSAI RP3-01 interface, this represents an extension of the RP3 (Reference Point 3) protocol for remote radio unit use. The BS can support multiple RRUs connected in chain, ring and tree-and-branch topologies, which makes the interface very flexible. Also, in order to minimize the number of connections to RRUs, the RP1 management plan, which includes Ethernet and frame clock bursts, is mapped into RP3 messages. This solution is an alternative to the design in which the radio module collocates with the BBM. Although in such a case the interface between the radio unit and the BBM becomes less complex, the transmitter power should be increased in order to compensate the feeder loss. For the proposed WiMAX BS block scheme, some improvements can be done starting from the proprieties of OBSAI RP3-01 interface. In this proposed BS split architecture, a BBM is connected to the two RRUs in order to have multiple transmit/ receive antennas for MIMO capabilities. The connection between the two RRUs is realized using a chain topology. In order to obtain a single point failure redundancy scheme, a second BBM connected to the two RRUs is required. Only one BBM will be active at the

Hub) modules and the AMC (Advanced Mezzanine Card) [6] modules.

**1. Introduction** 

different levels of the system.

**Split Architecture with MIMO** 

**Capabilities Support Based** 

**on OBSAI RP3-01 Interfaces** 

Cristian Anghel and Remus Cacoveanu

*University Politehnica of Bucharest,* 


### **MicroTCA Compliant WiMAX BS Split Architecture with MIMO Capabilities Support Based on OBSAI RP3-01 Interfaces**

Cristian Anghel and Remus Cacoveanu *University Politehnica of Bucharest, Romania* 

#### **1. Introduction**

18 Will-be-set-by-IN-TECH

76 Advanced Transmission Techniques in WiMAX

Tarokh, V., Jafarkhani, H. & Calderbank, A. R. (1999). Space - time block codes from orthogonal

Tellado, J. (2000). *Peak-to-average Power Reduction for Multicarrier modulation*, PhD thesis,

Tsiligkaridis, T. & Jones, D. L. (2010). PAPR reduction performance by active constellation

Wang, S.-H. & Li, C.-P. (2009). A low-complexity PAPR reduction scheme for SFBC

Zhang, H. & Goeckel, D. L. (2007). Peak power reduction in closed-loop MIMO-OFDM

Zhou, G. T. & Raich, R. (2004). Spectral analysis of polynomial nonlinearity with applications

to rf power amplifiers, *EURASIP Journal on Applied Signal Processing* 12: 1831–1840.

MIMO-OFDM systems, *IEEE Signal Proc. letters* 16: 941–945.

systems via mode reservation, *IEEE Commun. Lett.* 11: 583–586.

extension for diversity MIMO-OFDM systems, *Journal of Electrical and Computer*

designs, *IEEE Trans. Inf. Theory* 45: 1456–1467.

Stanford Univeristy, Stanford.

*Engineering* 2010: 5–9.

Modern mobile communication systems must fulfill more and more requirements received from the customers. This leads to an increase of complexity. The control part of the system becomes very important, a multi-level approach being needed. With respect to this, all BS (Base Stations) from a system are synchronized using GPS (Global Positioning System) or IEEE 1588 [1] standard, high speed synchronous interfaces are used between the BBM (Baseband Modules) and the RRU (Remote Radio Units), for example OBSAI (Open Base Station Architecture Initiative) [2, 3] or CPRI (Common Public Radio Interface) [4], and standard communication methods are provided between the control parts placed in different levels of the system.

This chapter describes the management and synchronization procedures for a WiMAX BS architecture compliant with MicroTCA standard (Micro Telecommunications Computing Architecture) [5]. The block scheme of such a BS for the case of a 3 sectors cell is presented. One can observe the main parts of the MicroTCA standard, i.e. the MCH (MicroTCA Carrier Hub) modules and the AMC (Advanced Mezzanine Card) [6] modules.

Referring now to the OBSAI RP3-01 interface, this represents an extension of the RP3 (Reference Point 3) protocol for remote radio unit use. The BS can support multiple RRUs connected in chain, ring and tree-and-branch topologies, which makes the interface very flexible. Also, in order to minimize the number of connections to RRUs, the RP1 management plan, which includes Ethernet and frame clock bursts, is mapped into RP3 messages. This solution is an alternative to the design in which the radio module collocates with the BBM. Although in such a case the interface between the radio unit and the BBM becomes less complex, the transmitter power should be increased in order to compensate the feeder loss. For the proposed WiMAX BS block scheme, some improvements can be done starting from the proprieties of OBSAI RP3-01 interface. In this proposed BS split architecture, a BBM is connected to the two RRUs in order to have multiple transmit/ receive antennas for MIMO capabilities. The connection between the two RRUs is realized using a chain topology. In order to obtain a single point failure redundancy scheme, a second BBM connected to the two RRUs is required. Only one BBM will be active at the

MicroTCA Compliant WiMAX BS Split Architecture with

module can also be present in a chassis, from redundancy reasons.

interface controllers, the base band and RF processing modules.

architecture elements, having only a support role.

Platform Management Interface) [7] standard.

distributed in the chassis and are described in Figure 2.1.

obtained fuctions.

**2.1 AMC modules** 

feature).

**2.2 MicroTCA Carrier** 

MIMO Capabilities Support Based on OBSAI RP3-01 Interfaces 79

The proposed architecture is a modular one, as one can see from Figure 2.1. It results in a very flexible solution, allowing a high diversity in AMC modules implementation and in

A MicroTCA chassis is made of 1 up to 12 AMC modules, which will realize together the system functionality. Then there is a MCH (MicroTCA Carrier Hub) module which represents the support for the implementation of the main system management functions. On the chassis there can be found also PM (Power Modules) modules used for power supplying, CU (Cooling Unit) modules used for temperature control at system level, interconnection elements between the modules or with the external inputs / outputs (Backplane, Faceplate) plus other mechanical elements and redundant modules. A second MCH module and a second CU

The AMC modules are the main components of a MicroTCA chassis, containing the elements which will provide the system processing functions. There can be listed here the microcontrollers, the digital signal processors, the routers, the memory blocks, the I/O

Initially, in AdvancedMC specifications, the AMC modules were defined as additional boards used for CB (Carrier Board) functionality extension. In MicroTCA, the AMC modules totally perform the system processing, while CB will be distributed between different

Due to the signal processing functions, the management tasks implemented on the AMC module are to be reduced as much as possible, in order to provide the maximum of the resources to the main process. This is the reason why these modules are controlled by a low level functionality entity called MMC (Module Management Controller). The set of functions this entity is performing is very simple so that it can be implemented on a low cost processor. The communication between MMC and a dedicated management entity at chassis level is made through IPMB-L (Intelligent Platform Management Bus - Local) interface, using a reduced set of requests/ confirmations specified in IPMI v2.0 (Intelligent

The IPMB-L connections are isolated between each others in order to avoid the case when

The most important advantage introduced by this standard is the possibility of introducing/ switching in the system any module without being required to stop the power supply (Hot Swap feature) or to make any other hardware/ software modification (Plug and Play

MC (MicroTCA Carrier) is the novelty proposed by MicroTCA and it represents the main board, as defined by AdvancedTCA. It is responsible for the power distribution, the interconnection and the IPMI management for the 12 AMC modules. MC components are

module issue is blocking the complete system, as in the case of bus topology.

time. There are also described the OBSAI RP3-01 Interfaces required for blocks interconnection. Note that on OBSAI RP3-01 interface of each RRU the same Transport and Application Layers serve the both Physical and Data Link Layers.

This chapter is organized as following: Section 2 describes briefly the MicroTCA standard and the most important elements of such an architecture. Section 3 proposes a simple and efficient method of synchronizing a WiMAX BS using the GPS signals. There are provided synchronization signals for the air interface, in order to avoid interferences with other BSs. Also there are obtained, based on this proposed method, synchronization signals used inside BS with the scope of aligning all the modules of the architecture, which is very important when split solution is adopted, i.e. not all the units are co-located in the same physical element. Finally, Section 4 proposes a new way of using OBSAI RP3-01 Interface in a WiMAX BS, this new implementation solution providing support for MIMO techniques and redundancy.

#### **2. MicroTCA standard – Overview**

The MicroTCA standard is created by PICMG (PCI Industrial Computer Manufacturers Group) and it defines the requirements of chassis hardware system. Such a system uses AMC (Advanced Mezzanine Card) modules interconnected by a board having a high speed interface on the backplane of the chassis. The standard defines the mechanical, electrical and management specific characteristics needed for supporting AMC standard compliant modules.

The described structure is a modular one. By the configuration and the interconnection of the AMC modules inside the chassis, a high variety of applications can be obtained. Besides this, the standard doesn't impose a certain physical configuration of the chassis and neither a mandatory signaling protocol for the backplane high speed interface. Instead, a set of communication and interconnection requirements is defined. This set of requirements should be available for any structure, providing this way a high compatibility between the equipments compliant with the standard.

Fig. 2.1 MicroTCA – Block scheme

The proposed architecture is a modular one, as one can see from Figure 2.1. It results in a very flexible solution, allowing a high diversity in AMC modules implementation and in obtained fuctions.

A MicroTCA chassis is made of 1 up to 12 AMC modules, which will realize together the system functionality. Then there is a MCH (MicroTCA Carrier Hub) module which represents the support for the implementation of the main system management functions. On the chassis there can be found also PM (Power Modules) modules used for power supplying, CU (Cooling Unit) modules used for temperature control at system level, interconnection elements between the modules or with the external inputs / outputs (Backplane, Faceplate) plus other mechanical elements and redundant modules. A second MCH module and a second CU module can also be present in a chassis, from redundancy reasons.

#### **2.1 AMC modules**

78 Advanced Transmission Techniques in WiMAX

time. There are also described the OBSAI RP3-01 Interfaces required for blocks interconnection. Note that on OBSAI RP3-01 interface of each RRU the same Transport and

This chapter is organized as following: Section 2 describes briefly the MicroTCA standard and the most important elements of such an architecture. Section 3 proposes a simple and efficient method of synchronizing a WiMAX BS using the GPS signals. There are provided synchronization signals for the air interface, in order to avoid interferences with other BSs. Also there are obtained, based on this proposed method, synchronization signals used inside BS with the scope of aligning all the modules of the architecture, which is very important when split solution is adopted, i.e. not all the units are co-located in the same physical element. Finally, Section 4 proposes a new way of using OBSAI RP3-01 Interface in a WiMAX BS, this new implementation solution providing support for MIMO techniques and redundancy.

The MicroTCA standard is created by PICMG (PCI Industrial Computer Manufacturers Group) and it defines the requirements of chassis hardware system. Such a system uses AMC (Advanced Mezzanine Card) modules interconnected by a board having a high speed interface on the backplane of the chassis. The standard defines the mechanical, electrical and management specific characteristics needed for supporting AMC standard compliant modules. The described structure is a modular one. By the configuration and the interconnection of the AMC modules inside the chassis, a high variety of applications can be obtained. Besides this, the standard doesn't impose a certain physical configuration of the chassis and neither a mandatory signaling protocol for the backplane high speed interface. Instead, a set of communication and interconnection requirements is defined. This set of requirements should be available for any structure, providing this way a high compatibility between the

Application Layers serve the both Physical and Data Link Layers.

**2. MicroTCA standard – Overview** 

equipments compliant with the standard.

Fig. 2.1 MicroTCA – Block scheme

The AMC modules are the main components of a MicroTCA chassis, containing the elements which will provide the system processing functions. There can be listed here the microcontrollers, the digital signal processors, the routers, the memory blocks, the I/O interface controllers, the base band and RF processing modules.

Initially, in AdvancedMC specifications, the AMC modules were defined as additional boards used for CB (Carrier Board) functionality extension. In MicroTCA, the AMC modules totally perform the system processing, while CB will be distributed between different architecture elements, having only a support role.

Due to the signal processing functions, the management tasks implemented on the AMC module are to be reduced as much as possible, in order to provide the maximum of the resources to the main process. This is the reason why these modules are controlled by a low level functionality entity called MMC (Module Management Controller). The set of functions this entity is performing is very simple so that it can be implemented on a low cost processor. The communication between MMC and a dedicated management entity at chassis level is made through IPMB-L (Intelligent Platform Management Bus - Local) interface, using a reduced set of requests/ confirmations specified in IPMI v2.0 (Intelligent Platform Management Interface) [7] standard.

The IPMB-L connections are isolated between each others in order to avoid the case when module issue is blocking the complete system, as in the case of bus topology.

The most important advantage introduced by this standard is the possibility of introducing/ switching in the system any module without being required to stop the power supply (Hot Swap feature) or to make any other hardware/ software modification (Plug and Play feature).

#### **2.2 MicroTCA Carrier**

MC (MicroTCA Carrier) is the novelty proposed by MicroTCA and it represents the main board, as defined by AdvancedTCA. It is responsible for the power distribution, the interconnection and the IPMI management for the 12 AMC modules. MC components are distributed in the chassis and are described in Figure 2.1.

MicroTCA Compliant WiMAX BS Split Architecture with

MCH module in the chassis for redundancy.

Fig. 2.3 MCH – Block scheme

synchronization at system level.

acts as MMC for the MCH.

protocols.

components, with a required precision.

MIMO Capabilities Support Based on OBSAI RP3-01 Interfaces 81

for all the other modules in the chassis. Having this important role, the MCH module is a critical point of the MicroTCA architecture, and for this reason it is required to have another

The MCH module represents the physical support for a set of control and management functions called CM (Carrier Manager), which is the main authority in MicroTCA Carrier. It has to deal with the power control, the AMC network connecting management, the IPMI control and management, the E-Keying functions, the Hot Swap functions and the




The logical link between the modules connected like this is made by the E-Keying function provided by the Carrier Manager. This function verifies that all the AMC units from a chassis are electrically compatible before giving the authorization to enter in the network.

The MCH structure and functionalities are based on the following modules:

The power distribution infrastructure

It provides and controls the power distribution for each AMC module. The standard indicates the existence of 3 functional aspects:


The PM units include also system surveillance functions required for the management part. The PM circuits should detect the units in the chassis, should monitor the parameters quality on each branch and should provide protection for overload. Part of these functions is made locally by a low level intelligence entity called EMMC (Enhanced Module Management Controller), but the system will be controlled by the Carrier Manager, which will compute the power budget based on the FRUI (Field Replaceable Unit Information) tables before validating the power distribution for AMC units. This process is similar with the AdvancedTCA one, and it is described in Figure 2.2.

Fig. 2.2 Power distribution infrastructure

Test infrastructure

It is represented by a JTAG switch called JSM (JTAG Switch Module). This allows the verification of the chassis, together with the modules inside, both in production period and in normal functioning period. As an alternative solution, serial asynchronous UART interfaces are implemented for the same testing scope.

#### **2.3 MicroTCA Carrier Hub**

This module combines the control and management infrastructure with the interconnection one, as depicted in Figure 2.3. It provides support for the 12 AMC units. Also, it is available for all the other modules in the chassis. Having this important role, the MCH module is a critical point of the MicroTCA architecture, and for this reason it is required to have another MCH module in the chassis for redundancy.

Fig. 2.3 MCH – Block scheme

80 Advanced Transmission Techniques in WiMAX

It provides and controls the power distribution for each AMC module. The standard

 the management supply MS providing 3.3V to each AMC. OS and MS are separated sub-systems in order to ensure the isolation between the processing

the distribution control logic DCL being responsible of the protection, isolation and

The PM units include also system surveillance functions required for the management part. The PM circuits should detect the units in the chassis, should monitor the parameters quality on each branch and should provide protection for overload. Part of these functions is made locally by a low level intelligence entity called EMMC (Enhanced Module Management Controller), but the system will be controlled by the Carrier Manager, which will compute the power budget based on the FRUI (Field Replaceable Unit Information) tables before validating the power distribution for AMC units. This process is similar with

It is represented by a JTAG switch called JSM (JTAG Switch Module). This allows the verification of the chassis, together with the modules inside, both in production period and in normal functioning period. As an alternative solution, serial asynchronous UART

This module combines the control and management infrastructure with the interconnection one, as depicted in Figure 2.3. It provides support for the 12 AMC units. Also, it is available

The power distribution infrastructure

indicates the existence of 3 functional aspects:

processes and the management ones.

the AdvancedTCA one, and it is described in Figure 2.2.

Fig. 2.2 Power distribution infrastructure

interfaces are implemented for the same testing scope.

Test infrastructure

**2.3 MicroTCA Carrier Hub** 

validation functions for each network branch

the operational supply OS providing 12V to each AMC

The MCH module represents the physical support for a set of control and management functions called CM (Carrier Manager), which is the main authority in MicroTCA Carrier. It has to deal with the power control, the AMC network connecting management, the IPMI control and management, the E-Keying functions, the Hot Swap functions and the synchronization at system level.

The MCH structure and functionalities are based on the following modules:


The logical link between the modules connected like this is made by the E-Keying function provided by the Carrier Manager. This function verifies that all the AMC units from a chassis are electrically compatible before giving the authorization to enter in the network.

MicroTCA Compliant WiMAX BS Split Architecture with

control the frequency generated by a local quart.

kind of requirements will be presented next in this chapter.

**3. WiMAX base station GPS based synchronization** 

obtaining high synchronization performances [9].

optic fiber.

station components.

**3.1 Introduction** 

MIMO Capabilities Support Based on OBSAI RP3-01 Interfaces 83



Having on one hand the WiMAX base station modules characteristics and on the other hand the MicroTCA architecture, one can identify that the CSM module has MCH specific functions, while the BBMs can be considered as being AMCs. Of course that besides these main modules, the power suppliers and the cooling units have to be added as WiMAX base

In communications systems using TDD (Time Division Duplex), appropriate time synchronization is critically important. In order to avoid inter-cell interference, all base stations must use the same timing reference. One solution to this problem is the Global Positioning System (GPS). The users can receive accurate time from atomic clocks and can generate themselves synchronization signals. Commonly the GPS receiver generates a Pulse per Second (PPS) signal and, optionally, a 10 MHz signal, phase synchronized with the PPS. All the transmission over the radio channel, both on downlink and uplink, should be synchronized with the PPS signal. The RP1 synchronization burst generator, called Clock Control Manager (CCM) shall provide frame timing and time stamping for each of the air interface systems independently. The quality of the PPS signal will dictate the periodicity of these synchronization bursts. Also, algorithms for maintaining the stability of the clock reference, which can be affected by the temperature variance or by aging, can be developed based on the PPS signal. It is obvious why the PPS jitter level is a critical parameter in

This document will describe the digital method used for PPS de-jittering and the VCXO

(Voltage Control Crystal Oscillator) oscillating frequency controlling algorithm.

system reference frequency generator. On the same time there are build in CCM (Control and Clock Module) the RP1 (Reference Point 1) synchronization signals used for OBSAI Interface and for Radio Interface. These FCBs (Frame Clock Burst) are sent time multiplexed to each BBM (Base Band Module) on a serial interface. On another serial interface all BBMs are receiving the system clock which will be used directly or to

#### **2.4 Base station components**

Based on the above description, the MicroTCA architecture totally fulfills the requirements of a mobile communication base station implementation. Properties like modularity, flexibility, high cooling capacity and low cost are supported by the standard. In addition, the MCH unit ensures an efficient control of the system elements and provides information required by the network high layers. The AMC modules have also an important role, being responsible of base band processing and RF processing that are required by a mobile communication system. A WiMAX base station example is described below based on the described MicroTCA architecture.

Fig. 2.4 WiMAX BS for a cell with 3 sectors

Figure 2.4 describes the WiMAX base station main components for the case of a cell with 3 sectors, each sector providing support for a certain level of diversity at transmission and reception. The main processing components are:


system reference frequency generator. On the same time there are build in CCM (Control and Clock Module) the RP1 (Reference Point 1) synchronization signals used for OBSAI Interface and for Radio Interface. These FCBs (Frame Clock Burst) are sent time multiplexed to each BBM (Base Band Module) on a serial interface. On another serial interface all BBMs are receiving the system clock which will be used directly or to control the frequency generated by a local quart.


Having on one hand the WiMAX base station modules characteristics and on the other hand the MicroTCA architecture, one can identify that the CSM module has MCH specific functions, while the BBMs can be considered as being AMCs. Of course that besides these main modules, the power suppliers and the cooling units have to be added as WiMAX base station components.

#### **3. WiMAX base station GPS based synchronization**

#### **3.1 Introduction**

82 Advanced Transmission Techniques in WiMAX

Based on the above description, the MicroTCA architecture totally fulfills the requirements of a mobile communication base station implementation. Properties like modularity, flexibility, high cooling capacity and low cost are supported by the standard. In addition, the MCH unit ensures an efficient control of the system elements and provides information required by the network high layers. The AMC modules have also an important role, being responsible of base band processing and RF processing that are required by a mobile communication system. A WiMAX base station example is described below based on the

MAC

PHY

MAC

RP1 FCBs

System Clock

PHY

MAC

PHY

Synchronization and control

BBM 2 (AMC 2)

Synchronization and control

BBM 3 (AMC 3)

Synchronization and control

Figure 2.4 describes the WiMAX base station main components for the case of a cell with 3 sectors, each sector providing support for a certain level of diversity at transmission and



BBM 1 (AMC 1)

OBSAI RP3-01

OBSAI RP3-01

OBSAI RP3-01

OBSAI RP3-01 Synchronization and control

OF

OF

OF

OBSAI RP3-01

Synchronization and control

> OBSAI RP3-01

Synchronization and control

RRU 1

RRU 2

RRU 3

Digital processing RF processing

Digital processing RF processing

Digital processing RF processing

**2.4 Base station components** 

described MicroTCA architecture.

PPS dejittering

CCM

Fig. 2.4 WiMAX BS for a cell with 3 sectors

extracted from the GPS system

reception. The main processing components are:

Reference frequency control

CSM (MCH)

GPS module

PPS

GPS antenna

In communications systems using TDD (Time Division Duplex), appropriate time synchronization is critically important. In order to avoid inter-cell interference, all base stations must use the same timing reference. One solution to this problem is the Global Positioning System (GPS). The users can receive accurate time from atomic clocks and can generate themselves synchronization signals. Commonly the GPS receiver generates a Pulse per Second (PPS) signal and, optionally, a 10 MHz signal, phase synchronized with the PPS.

All the transmission over the radio channel, both on downlink and uplink, should be synchronized with the PPS signal. The RP1 synchronization burst generator, called Clock Control Manager (CCM) shall provide frame timing and time stamping for each of the air interface systems independently. The quality of the PPS signal will dictate the periodicity of these synchronization bursts. Also, algorithms for maintaining the stability of the clock reference, which can be affected by the temperature variance or by aging, can be developed based on the PPS signal. It is obvious why the PPS jitter level is a critical parameter in obtaining high synchronization performances [9].

This document will describe the digital method used for PPS de-jittering and the VCXO (Voltage Control Crystal Oscillator) oscillating frequency controlling algorithm.

MicroTCA Compliant WiMAX BS Split Architecture with

Fig. 3.2 The bloc scheme of the digital part

*A. The de-jittering block* 

using equation 3.1:

where <sup>1</sup> 1..

*n*

Fig. 3.3 PPS jitter

2 *N*

MIMO Capabilities Support Based on OBSAI RP3-01 Interfaces 85

The PPS jitter characteristics are to be presented now. Figure 3.3 depicts the time instant value of the jitter. One can see from this figure that the PPS jitter is in the range 20 *ns* for the selected GPS receiver. Also is easy to observe that it does not have uniform distribution.

Figure 3.5 depicts the Allan deviation. For all of these measurements, it is assumed that the function *e*(*t*), representing the time error (the deviation from 1 second value), is sampled with *N* equally spaced samples, *ei* = *e*(*iτ*0), for *i* = (1*,* 2*, ...,N*), and with a sampling interval *τ*0 of 1 second. The observation interval, *τ*, is given by *τ* = *nτ*0. The Allan deviation is computed

> 0 2 2 2 0 1 <sup>1</sup> ( ) <sup>2</sup> 2 2


e

*ADEV n e ee n Nn*

PPS jitter

*N n*

*i*

0 100 200 300 400 500 600 700 800 900 1000

Time(sec)

<sup>2</sup> <sup>2</sup>

(3.1)

*i n in i*

For this reason a simple mean will not eliminate the jitter problem (see Figure 3.4).

#### **3.2 Clock reference controlling scheme**

The controlling scheme is a hybrid one, using both analog and digital elements. The scheme is depicted in Figure 3.1. In this application, the PPS input is sourced by a low cost GPS receiver called Resolution T, produced by Trimble.

Fig. 3.1 Controlling scheme

The scheme works as follows: the Field Programmable Gate Array (FPGA), which is a XC3S500E chip representing a Spartan 3E family member produced by Xilinx, increments a counter value on every *Fref* rising edge and resets this counter on every PPS pulse. Let's consider the nominal frequency *<sup>n</sup> Fref* and the counter value at a time instant *count val* \_ . When a new PPS pulse is received from the GPS module, before the counter reset, his value is stored and compared with *<sup>n</sup> Fref* . If the two values are not equal, the digital block computes a digital command *CMDd* that is converted into a voltage level by a Digital to Analog Converter (DAC). The analog command *CMDa* controls the VCXO and the value of *Fref* is changed accordingly.

As it was mentioned before, the PPS jitter can produce VCXO commands that are unnecessary or imprecise. This is the reason way the PPS signal from the GPS receiver is passed through a digital de-jittering block before it is used by the controlling algorithm and by the CCM. The bloc scheme of the digital part of the structure described in Figure 3.1 is depicted in Figure 3.2.

Fig. 3.2 The bloc scheme of the digital part

#### *A. The de-jittering block*

84 Advanced Transmission Techniques in WiMAX

The controlling scheme is a hybrid one, using both analog and digital elements. The scheme is depicted in Figure 3.1. In this application, the PPS input is sourced by a low cost GPS

The scheme works as follows: the Field Programmable Gate Array (FPGA), which is a XC3S500E chip representing a Spartan 3E family member produced by Xilinx, increments a counter value on every *Fref* rising edge and resets this counter on every PPS pulse. Let's consider the nominal frequency *<sup>n</sup> Fref* and the counter value at a time instant *count val* \_ . When a new PPS pulse is received from the GPS module, before the counter reset, his value is stored and compared with *<sup>n</sup> Fref* . If the two values are not equal, the digital block computes a digital command *CMDd* that is converted into a voltage level by a Digital to Analog Converter (DAC). The analog command *CMDa* controls the VCXO and the value of *Fref* is

As it was mentioned before, the PPS jitter can produce VCXO commands that are unnecessary or imprecise. This is the reason way the PPS signal from the GPS receiver is passed through a digital de-jittering block before it is used by the controlling algorithm and by the CCM. The bloc scheme of the digital part of the structure described in Figure 3.1 is

**3.2 Clock reference controlling scheme** 

Fig. 3.1 Controlling scheme

changed accordingly.

depicted in Figure 3.2.

receiver called Resolution T, produced by Trimble.

The PPS jitter characteristics are to be presented now. Figure 3.3 depicts the time instant value of the jitter. One can see from this figure that the PPS jitter is in the range 20 *ns* for the selected GPS receiver. Also is easy to observe that it does not have uniform distribution. For this reason a simple mean will not eliminate the jitter problem (see Figure 3.4).

Figure 3.5 depicts the Allan deviation. For all of these measurements, it is assumed that the function *e*(*t*), representing the time error (the deviation from 1 second value), is sampled with *N* equally spaced samples, *ei* = *e*(*iτ*0), for *i* = (1*,* 2*, ...,N*), and with a sampling interval *τ*0 of 1 second. The observation interval, *τ*, is given by *τ* = *nτ*0. The Allan deviation is computed using equation 3.1:

$$ADDV(n\tau\_0) = \sqrt{\frac{1}{2n^2\tau\_0^2} \left(N - 2n\right)} \sum\_{i=1}^{N-2n} \left(e\_{i+2n} - 2e\_{i+n} + e\_i\right)^2 \tag{3.1}$$
 where  $n \in \left(1, \left\lceil \frac{N-1}{2} \right\rceil\right)$ 

$$\begin{aligned} \underbrace{\frac{1}{n} \left(\left\lfloor \frac{N}{2} \right\rfloor\right)^2 \left(\left\lfloor \frac{N}{2} \right\rfloor\right)^2 \left(\left\lfloor \frac{N}{2} \right\rfloor\right)}\_{\text{\tiny the \text{\tiny D}}} \left(\left\lfloor \frac{N}{2} \right\rfloor\right) \left(\left\lfloor \frac{N}{2} \right\rfloor\right) \left(\left\lfloor \frac{N}{2} \right\rfloor\right) \left(\left\lfloor \frac{N}{2} \right\rfloor\right)^2 \dots \tag{3.2} \\ \underset{\left\lfloor \frac{N}{2} \right\rfloor}{\text{\tiny the \text{\tiny D}}} \left(\left\lfloor \frac{N}{2} \right\rfloor\right) \left(\left\lfloor \frac{N}{2} \right\rfloor\right) \left(\left\lfloor \frac{N}{2} \right\rfloor\right) \left(\left\lfloor \frac{N}{2} \right\rfloor\right) \dots \tag{3.3} \end{aligned} \tag{3.4}$$

MicroTCA Compliant WiMAX BS Split Architecture with

used as the Kalman filter input.

de-jittering algorithm is as follows:

MIMO Capabilities Support Based on OBSAI RP3-01 Interfaces 87

If the oscillator frequency is *<sup>n</sup> Fref* then *DIFF(n)* will reflect only the PPS jitter. If not, the *DIFF(n)* will contain the frequency deviation also. These values, computed every second, are

Using the notations *Q* for process variance and *R* for estimate of measurement variance, the

*Initialization*

(3.3)

(3.4)

 

The *DIFFdj* signal from Figure 3.2 is the filter output, i.e. *x n* ( ) , and it is used to compute the digital command for the VCXO. Also, the de-jittering block provides a de-jittered PPS pulse,

Some times, due to the lack of visibility, the GPS receiver might not transmit the PPS pulse. This situation should be detected by the State Controller by expecting the PPS pulse within a time window. This window depends on the oscillator stability. If the oscillator has a 25*ppm* variation within the temperature range and a nominal frequency of 153.6 MHz, then the maximum delay of the PPS pulse can be 153.6 6 25 6 3840 *ex e* clock periods, i.e. the PPS pulse can be found after the previous one at 153.600.000±3840 clock periods. If it is

The Finite State Machine (FSM) of the synchronization block is depicted in Figure 3.6. The

TRAINING: when the synchronization block starts the Kalman filtering and waits *TTR*

NORMAL: when the synchronization block works based on PPS pulses received form

*K n DIFF n x n Pn Kn P n*

( 1) ( ) ( 1) ( 1) 1 ( 1) ( 1)

 

( 1) ( 1) / ( 1)

*Kn P n P n R*

(1) 0; (1) 1;

 

*x P*

1 : ( )

*for n N Time update x n xn P n Pn Q Measurement update*

 

( 1) ( ) ( 1) ( ) ( )

 

( 1) ( 1)

*xn x n*

not so, the State Controller block confirms the absence of the PPS pulse.

IDLE: when the synchronization block waits for the first PPS

seconds in order to obtain a stable output

*end*

denoted *PPSdj* which should have a 1 second period.

B. *State Controller*

four possible states are:

the GPS module

Fig. 3.4 PPS jitter mean

Fig. 3.5 PPS jitter Allan deviation

The slope of *ADEV*( ) is <sup>1</sup> , which corresponds to white noise phase modulation and flicker phase modulation [1].

The de-jittering block contains a discreet-time Kalman filter. We will consider a particular algorithm of one-dimension Kalman filter intended for frequency estimation only in oscillators if GPS timing signals are used as the reference ones [10]. As it was mentioned before, on every PPS pulse we compute:

$$\text{DIFF}(n) = \text{count\\_val}(n) - F\_{\text{ref}}^n \tag{3.2}$$

If the oscillator frequency is *<sup>n</sup> Fref* then *DIFF(n)* will reflect only the PPS jitter. If not, the *DIFF(n)* will contain the frequency deviation also. These values, computed every second, are used as the Kalman filter input.

Using the notations *Q* for process variance and *R* for estimate of measurement variance, the de-jittering algorithm is as follows:

$$\begin{array}{c} \text{Initialization} \\ \tilde{\mathbf{x}}(1) = 0; \\ P(1) = 1; \end{array} \tag{3.3}$$
  $for \, n = 1:N$   $(Time \, update)$  
$$\begin{array}{c} \tilde{\mathbf{x}}\_{-}(n+1) = \tilde{\mathbf{x}}(n) \\ P\_{-}(n+1) = P(n) + Q \end{array}$$

$$\begin{aligned} P\_{-(n+1)} &= P\_{(n)} \tau \otimes \\ &\quad \text{(Measurement update)}\\ K(n+1) &= P\_{-(n+1)} / \left( P\_{-(n+1)} + R \right) \\ &\quad \tilde{\mathbf{x}}(n+1) = \tilde{\mathbf{x}}\_{-}(n+1) + \\ &\quad K(n+1) (DIFF(n) - \tilde{\mathbf{x}}\_{-}(n+1)) \\ &\quad P(n+1) = \left( 1 - K(n+1) \right) P\_{-}(n+1) \\ &\quad \text{end} \end{aligned} \tag{3.4}$$

The *DIFFdj* signal from Figure 3.2 is the filter output, i.e. *x n* ( ) , and it is used to compute the digital command for the VCXO. Also, the de-jittering block provides a de-jittered PPS pulse, denoted *PPSdj* which should have a 1 second period.

#### B. *State Controller*

86 Advanced Transmission Techniques in WiMAX

<sup>14</sup> x 10-9 jitter mean

mean over 64 seconds

<sup>0</sup> <sup>2000</sup> <sup>4000</sup> <sup>6000</sup> <sup>8000</sup> <sup>10000</sup> <sup>12000</sup> <sup>14000</sup> <sup>16000</sup> <sup>18000</sup> -6

Time(sec)

Allan deviation

<sup>100</sup> <sup>101</sup> <sup>102</sup> <sup>103</sup> 10-12

Time(sec)

The de-jittering block contains a discreet-time Kalman filter. We will consider a particular algorithm of one-dimension Kalman filter intended for frequency estimation only in oscillators if GPS timing signals are used as the reference ones [10]. As it was mentioned

, which corresponds to white noise phase modulation and

() \_ () *<sup>n</sup> DIFF n count val n F ref* (3.2)

10-11

before, on every PPS pulse we compute:

 is <sup>1</sup> 

10-10

ADEV

Fig. 3.5 PPS jitter Allan deviation

The slope of *ADEV*( )

flicker phase modulation [1].

10-9

10-8

Fig. 3.4 PPS jitter mean

me

Some times, due to the lack of visibility, the GPS receiver might not transmit the PPS pulse. This situation should be detected by the State Controller by expecting the PPS pulse within a time window. This window depends on the oscillator stability. If the oscillator has a 25*ppm* variation within the temperature range and a nominal frequency of 153.6 MHz, then the maximum delay of the PPS pulse can be 153.6 6 25 6 3840 *ex e* clock periods, i.e. the PPS pulse can be found after the previous one at 153.600.000±3840 clock periods. If it is not so, the State Controller block confirms the absence of the PPS pulse.

The Finite State Machine (FSM) of the synchronization block is depicted in Figure 3.6. The four possible states are:


MicroTCA Compliant WiMAX BS Split Architecture with

computed as:

Fig. 3.7 Control Algorithm

block was in NORMAL state.

of the PPS pulse.

MIMO Capabilities Support Based on OBSAI RP3-01 Interfaces 89

153.6 6 32 6 <sup>1</sup> 0.075 2 13

The *CMDi* signal used as feedback for the controlling loop is selected as described in Figure 3.7. The *DIFF* and *DIFFdj* are expressed in clock periods per second and so the DAC value is

*<sup>f</sup> Hz Hz* (3.5)

*CMD k CMD k CMD dd i* 1 13 (3.6)

The starting value is the central level of the DAC range, i.e. 215. In order to obtain a faster convergence, the starting value might be a DAC value saved when the synchronization

When the State Controller indicates IDLE or TRAINING the oscillator is controlled directly with the measured frequency deviation, in order to achieve fast convergence. In NORMAL state, the Kalman output is used for jitter reduction. The *DIFFdj* signal has a floating point format, so that frequency corrections less than 1 Hz can be produced. Also the mean of the *Nm* last values of *DIFFdj* signal is computed. The mean value is used when the State Controller is in HOLD OVER state and no valid *DIFFdj* values are received. Also this value is maintained for *THO-N* seconds when the synchronizations state returns from HOLD OVER state to NORMAL state, in order not to produce de-synchronization due to the new position

The values of the Control Algorithm parameters are given in Table 3.2.

16

*ex e*

 HOLD OVER: when the PPS pulse is not received from the GPS module and the synchronization block works based on local PPS pulse.

Fig. 3.6 FSM for synchronization block

After the first PPS received, the synchronization block switches from IDLE to TRAINING. In TRAINING, a counter called *countTR* is incremented on every PPS pulse. If the counter value equals the *TTR* parameter, then a transition is made in NORMAL state. Else, if the block declares the absence of the PPS pulse and the counter value is less than *TTR* then the new state becomes IDLE.

When the FSM is in NORMAL state and the PPS is declared to be absent, a transition to HOLD OVER state is made. In this state, from the last PPS pulse received, a counter is started in order to generate an internal PPS pulse. Also, a counter called *countHO* is incremented on every local PPS pulse, counting the number of successive absent external PPS pulses. If this counter reaches *THO* parameter the synchronization bloc state becomes IDLE. Else, if a new PPS pulse is detected before the counter reaches the *THO* value, the synchronization block returns in NORMAL state.

The values of the FSM parameters are given in Table 3.1.


Table 3.1 FSM parameters

#### C. *Control Algorithm*

The Control Algorithm block receives the Kalman output and the state of the synchronization block. It also receives the Kalman input, as one can see from Figure 3.2. The control algorithm should compute the VCXO command. The DAC has a 16-bit input, so <sup>16</sup> 2 values are used to control the range of the VCXO. For a measured frequency deviation of 16*ppm* , result a control step of:

HOLD OVER: when the PPS pulse is not received from the GPS module and the

After the first PPS received, the synchronization block switches from IDLE to TRAINING. In TRAINING, a counter called *countTR* is incremented on every PPS pulse. If the counter value equals the *TTR* parameter, then a transition is made in NORMAL state. Else, if the block declares the absence of the PPS pulse and the counter value is less than *TTR* then the

When the FSM is in NORMAL state and the PPS is declared to be absent, a transition to HOLD OVER state is made. In this state, from the last PPS pulse received, a counter is started in order to generate an internal PPS pulse. Also, a counter called *countHO* is incremented on every local PPS pulse, counting the number of successive absent external PPS pulses. If this counter reaches *THO* parameter the synchronization bloc state becomes IDLE. Else, if a new PPS pulse is detected before the counter reaches the *THO* value, the

> *Parameter Value Unit*  TTR 192 sec **THO** Depending on VCXO sec

The Control Algorithm block receives the Kalman output and the state of the synchronization block. It also receives the Kalman input, as one can see from Figure 3.2. The control algorithm should compute the VCXO command. The DAC has a 16-bit input, so <sup>16</sup> 2 values are used to control the range of the VCXO. For a measured frequency deviation of

synchronization block works based on local PPS pulse.

Fig. 3.6 FSM for synchronization block

synchronization block returns in NORMAL state.

The values of the FSM parameters are given in Table 3.1.

new state becomes IDLE.

Table 3.1 FSM parameters

16*ppm* , result a control step of:

C. *Control Algorithm*

$$Af = \frac{153.6 \, e \, e \, 32 \, e - 6}{2^{16}} = 0.075 \, Hz \approx \frac{1}{13} \, Hz \tag{3.5}$$

The *CMDi* signal used as feedback for the controlling loop is selected as described in Figure 3.7. The *DIFF* and *DIFFdj* are expressed in clock periods per second and so the DAC value is computed as:

Fig. 3.7 Control Algorithm

$$\text{CMD}\_d(k) = \text{CMD}\_d(k-1) - 1\text{BMD}\_i \tag{3.6}$$

The starting value is the central level of the DAC range, i.e. 215. In order to obtain a faster convergence, the starting value might be a DAC value saved when the synchronization block was in NORMAL state.

When the State Controller indicates IDLE or TRAINING the oscillator is controlled directly with the measured frequency deviation, in order to achieve fast convergence. In NORMAL state, the Kalman output is used for jitter reduction. The *DIFFdj* signal has a floating point format, so that frequency corrections less than 1 Hz can be produced. Also the mean of the *Nm* last values of *DIFFdj* signal is computed. The mean value is used when the State Controller is in HOLD OVER state and no valid *DIFFdj* values are received. Also this value is maintained for *THO-N* seconds when the synchronizations state returns from HOLD OVER state to NORMAL state, in order not to produce de-synchronization due to the new position of the PPS pulse.

The values of the Control Algorithm parameters are given in Table 3.2.

MicroTCA Compliant WiMAX BS Split Architecture with


> -8 -6 -4 -2 0 2 4 6 8 10

Fig. 3.11 DIFFdj values in clock periods

DIFFdj

DIFF

Fig. 3.10 DIFF values in clock periods

of the jitter is much lower.

MIMO Capabilities Support Based on OBSAI RP3-01 Interfaces 91

Figure 3.10 depicts the deviation measured at the input and of the Kalman filter, while Figure 3.11 depicts the deviation measured at the output of the Kalman filter One can see that the initial deviation is the jitter level ( 20 3 *nsxFref* clock periods) plus some temperature randomly added deviation. At the output of the de-jittering structure, the level

Input of Kalman filter

<sup>0</sup> <sup>2000</sup> <sup>4000</sup> <sup>6000</sup> <sup>8000</sup> <sup>10000</sup> <sup>12000</sup> <sup>14000</sup> <sup>16000</sup> -10

<sup>0</sup> <sup>2000</sup> <sup>4000</sup> <sup>6000</sup> <sup>8000</sup> <sup>10000</sup> <sup>12000</sup> <sup>14000</sup> <sup>16000</sup> -10

Figure 3.12 and Figure 3.13 depicts the deviation measured at the input and output of the Kalman filter while a frequency deviation of 1 Hz per second was added for 500 successive seconds and subtracted for the next 500 seconds. One can observe that the *DIFFdj* signal indicates the need of about 1 Hz reduction of the oscillating frequency for the first 500

seconds and then the need of about 1 Hz adding for the next 500 seconds.

Time(sec)

Time(sec)

Output of Kalman filter


Table 3.2 Control Algorithm parameters

#### **3.3 Experimental results**

The VCXO oscillating frequency is affected by the temperature variations. The controlling algorithm should provide commands fast enough not to accumulate frequency deviations. This problem is observed at the start up too. Even if the temperature is stable, the oscillator is not in a stable thermal state and the algorithm should provide frequency corrections. Figure 3.8 depicts the mean of the DAC commands, considering 0x8000 value as the reference. One can see that at start-up the oscillator has a significant frequency drift and although the temperature variation, depicted in Figure 3.9, is not important, the frequency deviation is about 10Hz per 9000 seconds.

Fig. 3.8 Mean of frequency deviation

Fig. 3.9 Temperature variation

*Parameter Value Unit*

**THO-N** 2 sec

The VCXO oscillating frequency is affected by the temperature variations. The controlling algorithm should provide commands fast enough not to accumulate frequency deviations. This problem is observed at the start up too. Even if the temperature is stable, the oscillator is not in a stable thermal state and the algorithm should provide frequency corrections. Figure 3.8 depicts the mean of the DAC commands, considering 0x8000 value as the reference. One can see that at start-up the oscillator has a significant frequency drift and although the temperature variation, depicted in Figure 3.9, is not important, the frequency

<sup>0</sup> <sup>1000</sup> <sup>2000</sup> <sup>3000</sup> <sup>4000</sup> <sup>5000</sup> <sup>6000</sup> <sup>7000</sup> <sup>8000</sup> <sup>9000</sup> <sup>10000</sup> <sup>372</sup>

Time(sec)

<sup>0</sup> <sup>1000</sup> <sup>2000</sup> <sup>3000</sup> <sup>4000</sup> <sup>5000</sup> <sup>6000</sup> <sup>7000</sup> <sup>8000</sup> <sup>9000</sup> <sup>10000</sup> 35.6

Time(sec)

Nm 128

Table 3.2 Control Algorithm parameters

deviation is about 10Hz per 9000 seconds.

374

35.8

36

36.2

Temperature(oC)

Fig. 3.9 Temperature variation

36.4

36.6

36.8

37

376

378

Frequency correction(Hz)

Fig. 3.8 Mean of frequency deviation

380

382

384

**3.3 Experimental results** 

Fig. 3.10 DIFF values in clock periods

Fig. 3.11 DIFFdj values in clock periods

Figure 3.12 and Figure 3.13 depicts the deviation measured at the input and output of the Kalman filter while a frequency deviation of 1 Hz per second was added for 500 successive seconds and subtracted for the next 500 seconds. One can observe that the *DIFFdj* signal indicates the need of about 1 Hz reduction of the oscillating frequency for the first 500 seconds and then the need of about 1 Hz adding for the next 500 seconds.

MicroTCA Compliant WiMAX BS Split Architecture with

standards, such as WCDMA, GSM/EDGE, CDMA2000 and 802.16.

should be increased in order to compensate the feeder loss.

**OBSAI RP3-01 Interface** 

**4.2 RP3-01 protocol stack** 

*A. Physical Layer* 

concern this study. *B. Data Link Layer* 

protocol with fixed length messages.

**4.1 Introduction** 

MIMO Capabilities Support Based on OBSAI RP3-01 Interfaces 93

**4. WiMAX base station architecture with MIMO capabilities support based on** 

More and more companies try to provide full solutions when it comes to wireless telecommunications systems. But unfortunately this concept, called ecosystem, is not always an easy task to realize. The costs are quite large and, due to system high complexity, the development time is also very long. That means it is very possible that parts of a system can be made by different vendors. The interconnection between different parts should be made by standard interfaces. The usage of third party intellectual property solution reduces the compatibility area. This is the reason why standards as Common Public Radio Interface (CPRI) and Open Base Station Architecture Initiative (OBSAI) were developed. The OBSAI RP3-01 interface permits the transport of data corresponding to different communications

The OBSAI RP3-01 interface represents an extension of the Reference Point 3 protocol for remote radio unit use. The BS can support multiple RRUs connected in chain, ring and treeand-branch topologies, which makes the interface very flexible. Also, in order to minimize the number of connections to RRUs, the RP1 management plan [3], which includes Ethernet and frame clock bursts, is mapped into RP3 messages. This solution is an alternative to the design in which the radio module collocates with the BBM. Although in such a case the interface between the radio unit and the BBM becomes less complex, the transmitter power

This section describes the RP3-01 protocol stack and the corresponding parameters, presents the synchronization procedure. Then a BS split architecture is proposed and the implementation solutions for the most important interface layers components are provided, together with the implementation results obtained when a XC4VFX60 FPGA is targeted.

The RP3-01 interface is a high spe4ed serial interface for both up link and down link data and control transfer. The protocol stack is based on a packet concept using a layered

The transmitter Physical Layer is responsible for the 8b10b encoding, which provides a mechanism for clock recovery, and data serialization. At receiver, the mirrored functions are applied. The Physical Layer can be implemented by a dedicated device, such as an XGMII transceiver [11], or by an internal FPGA component, such as RocketIO transceiver from Xilinx Virtex 5 family [12], when a hardware implementation is considered. The supported rates are 768 Mbps, 1536 Mbps, 3072 Mbps and 6144Mbps. The 6144 Mbps rate does not

The Data Link Layer contains the frame builder and the link synchronization unit. The frame builder receives from superior layers the data and control messages and generates, according to the transmitter rate, the RP3 frame. The data or control message has a fixed 19

 Fig. 3.12 DIFF values in clock periods

Fig. 3.13 DIFFdj values in clock periods

The results presented in this section were obtained using a hardware platform compliant with Figure 3.1. The inputs and the outputs of the algorithm were transferred to the PC using a UART interface. The pictures were obtained using Matlab.

#### **3.4 Conclusions**

This section presented a digital method of reducing the jitter level of the PPS signal generated by a GPS receiver. Also a controlling algorithm of a VCXO oscillating frequency was described. The results indicated that the frequency correction was applied only when the thermal state of the oscillator was not stable. False corrections due to the PPS jitter were almost completely eliminated.

#### **4. WiMAX base station architecture with MIMO capabilities support based on OBSAI RP3-01 Interface**

#### **4.1 Introduction**

92 Advanced Transmission Techniques in WiMAX

Input of Kalman filter

<sup>5000</sup> <sup>5200</sup> <sup>5400</sup> <sup>5600</sup> <sup>5800</sup> <sup>6000</sup> <sup>6200</sup> <sup>6400</sup> <sup>6600</sup> <sup>6800</sup> <sup>7000</sup> -10

Time(sec)

Output of Kalman filter

5000 5500 6000 6500 7000

The results presented in this section were obtained using a hardware platform compliant with Figure 3.1. The inputs and the outputs of the algorithm were transferred to the PC

This section presented a digital method of reducing the jitter level of the PPS signal generated by a GPS receiver. Also a controlling algorithm of a VCXO oscillating frequency was described. The results indicated that the frequency correction was applied only when the thermal state of the oscillator was not stable. False corrections due to the PPS jitter were

Time(sec)

Fig. 3.12 DIFF values in clock periods



using a UART interface. The pictures were obtained using Matlab.

Fig. 3.13 DIFFdj values in clock periods

**3.4 Conclusions** 

almost completely eliminated.

DIFFdj

DIFF

More and more companies try to provide full solutions when it comes to wireless telecommunications systems. But unfortunately this concept, called ecosystem, is not always an easy task to realize. The costs are quite large and, due to system high complexity, the development time is also very long. That means it is very possible that parts of a system can be made by different vendors. The interconnection between different parts should be made by standard interfaces. The usage of third party intellectual property solution reduces the compatibility area. This is the reason why standards as Common Public Radio Interface (CPRI) and Open Base Station Architecture Initiative (OBSAI) were developed. The OBSAI RP3-01 interface permits the transport of data corresponding to different communications standards, such as WCDMA, GSM/EDGE, CDMA2000 and 802.16.

The OBSAI RP3-01 interface represents an extension of the Reference Point 3 protocol for remote radio unit use. The BS can support multiple RRUs connected in chain, ring and treeand-branch topologies, which makes the interface very flexible. Also, in order to minimize the number of connections to RRUs, the RP1 management plan [3], which includes Ethernet and frame clock bursts, is mapped into RP3 messages. This solution is an alternative to the design in which the radio module collocates with the BBM. Although in such a case the interface between the radio unit and the BBM becomes less complex, the transmitter power should be increased in order to compensate the feeder loss.

This section describes the RP3-01 protocol stack and the corresponding parameters, presents the synchronization procedure. Then a BS split architecture is proposed and the implementation solutions for the most important interface layers components are provided, together with the implementation results obtained when a XC4VFX60 FPGA is targeted.

#### **4.2 RP3-01 protocol stack**

The RP3-01 interface is a high spe4ed serial interface for both up link and down link data and control transfer. The protocol stack is based on a packet concept using a layered protocol with fixed length messages.

#### *A. Physical Layer*

The transmitter Physical Layer is responsible for the 8b10b encoding, which provides a mechanism for clock recovery, and data serialization. At receiver, the mirrored functions are applied. The Physical Layer can be implemented by a dedicated device, such as an XGMII transceiver [11], or by an internal FPGA component, such as RocketIO transceiver from Xilinx Virtex 5 family [12], when a hardware implementation is considered. The supported rates are 768 Mbps, 1536 Mbps, 3072 Mbps and 6144Mbps. The 6144 Mbps rate does not concern this study.

#### *B. Data Link Layer*

The Data Link Layer contains the frame builder and the link synchronization unit. The frame builder receives from superior layers the data and control messages and generates, according to the transmitter rate, the RP3 frame. The data or control message has a fixed 19

MicroTCA Compliant WiMAX BS Split Architecture with

format of a data message payload field is shown in Figure 4.2.

*D. Application Layer* 

Fig. 4.2 802.16 payload mapping

**4.3 RP3-01 synchronization procedure** 

MIMO Capabilities Support Based on OBSAI RP3-01 Interfaces 95

Another block of the Transport Layer is the Message Multiplexer/ Demultiplexer. It performs time interleaving/ deinterleaving of messages from *N* RP3 input links into one RP3 output link. Several multiplexing/ demultiplexing tables are defined as functions of the

The Application Layer builds the data messages. It maps data and control information into the message payload and attaches the message header. The payload is represented by concatenation of signal samples in the baseband. For the case of the 802.16 air interface, the

All transfers over the bus are performed over paths. A path represents a connection between a source node and a destination node. The connection is made by a set of bus links defined by the routing tables. Physically, a path consists of a set of message slots per MF. Paths are defined before bus initialization and they remain fixed during operation, i.e. the transfer between two nodes is made on the same bus links using the same message slots. For each path, a message transmission rule is applied. There are two types of rules: mandatory low level rules, using modulo computation over message slot counters and optionally high level rules, using the Bit Map (BM) concept. In the second case each MG is formed as groups of

messages from baseband channels. Each group may contain Ethernet frames as well.

The PR1 Frame Clock Bursts (FCB) is mapped into RP3 messages.

In frame based communications systems, appropriate frequency/ time synchronization is critically important. In order to avoid inter-cell interference, all base stations must use the same frequency/ time reference. The transmission over the radio channel, both on downlink and uplink, should be synchronized with the same reference. The RP1 synchronization burst generator, located at Clock Control Manager (CCM) level, shall provide frame timing and time stamping for each of the air interface standard independently, based on that reference.

The synchronization algorithm uses several inputs and is realized based on information collected both at BBM and RRU. The first one is the propagation delay (PD) between BBM and RRU. This time interval is measured on BBM side using special message transmission called Round Trip Time (RTT) measurement message. The second input is the receiving time of the RP1 FCB. This interval is measured from the beginning of the last MF until the last bit of the FCB using a counter called *C1*. After measuring this time interval, the Frame

number of input links and their corresponding rates and the rate of the output link.

bytes length. The message format contains 4 fields. The first one is the Address field on 13 bits, the second one is the Type field on 5 bits, the third one is the Time Stamp field on 6 bits and the last one is the Payload field on 128 bits.

The duration of RP3-01 Master Frame (MF) is fixed to 10 ms. This length corresponds to *i* x *N\_MG* Master Groups (MG), where *i* is selected accordingly to the transfer rate, i.e. 1 for 768 Mbps, 2 for 1536 Mbps and 4 for 3072 Mbps. A MG consists of *M\_MG* messages and *K\_MG* idle bytes (special characters). Figure 4.1 presents an example of MF corresponding to 802.16 standard [13], when 768 Mbps rate is used. Also one can observe the values of the specified parameters.

Fig. 4.1 MF for 802.16 air interface standard for the 768Mbps line rate

For the example described in Figure 4.1, the transfer rate can be computed as follows: first we compute the maximum number of bytes per frame, and then, having in mind the 8b10b line encoder and the MF duration, we calculate the transfer rate, as in (4.1). In order to generate the MF, the frame builder uses two counters: for data messages and for control messages.

$$\text{Rate} = \text{i} \cdot \text{N}\_{\text{\\_}} \cdot \text{MG} \text{(M\\_} \text{\_MG} \cdot \text{19} + \text{K\\_} \text{\_MG} \cdot \text{1)} \cdot \text{1c\\$} \tag{4.1}$$

The link synchronization unit contains transmit and receive Finite State Machine (FSM). Both the transmitter and receiver FSMs contribute to physical and logical link synchronization. The physical synchronization is based on special characters, K28.5 and K28.7, which mark the end of message groups and MF, while the logical synchronization is based on fixed MF structure.

#### *C. Transport Layer*

The Transport Layer is responsible for the end-to-end delivery of the messages, which could be simply routing of messages. The routing is based on the first 13 bits of the message, which represent the address field. The first 8 bits of the address represent the Node address and the other 5 bits represent the Sub-node address. These fields are used in a hierarchical addressing scheme where the first field identifies the bus nod and the second one selects the corresponding module from the device. The message routing is made based on a Routing Table which indicates the correspondence between the used addresses and the node ports. The table content is defined by the initial configuration procedure of the interface.

Another block of the Transport Layer is the Message Multiplexer/ Demultiplexer. It performs time interleaving/ deinterleaving of messages from *N* RP3 input links into one RP3 output link. Several multiplexing/ demultiplexing tables are defined as functions of the number of input links and their corresponding rates and the rate of the output link.

#### *D. Application Layer*

94 Advanced Transmission Techniques in WiMAX

bytes length. The message format contains 4 fields. The first one is the Address field on 13 bits, the second one is the Type field on 5 bits, the third one is the Time Stamp field on 6 bits

The duration of RP3-01 Master Frame (MF) is fixed to 10 ms. This length corresponds to *i* x *N\_MG* Master Groups (MG), where *i* is selected accordingly to the transfer rate, i.e. 1 for 768 Mbps, 2 for 1536 Mbps and 4 for 3072 Mbps. A MG consists of *M\_MG* messages and *K\_MG* idle bytes (special characters). Figure 4.1 presents an example of MF corresponding to 802.16 standard [13], when 768 Mbps rate is used. Also one can observe the values of the specified

For the example described in Figure 4.1, the transfer rate can be computed as follows: first we compute the maximum number of bytes per frame, and then, having in mind the 8b10b line encoder and the MF duration, we calculate the transfer rate, as in (4.1). In order to generate the MF, the frame builder uses two counters: for data messages and for control

The link synchronization unit contains transmit and receive Finite State Machine (FSM). Both the transmitter and receiver FSMs contribute to physical and logical link synchronization. The physical synchronization is based on special characters, K28.5 and K28.7, which mark the end of message groups and MF, while the logical synchronization is

The Transport Layer is responsible for the end-to-end delivery of the messages, which could be simply routing of messages. The routing is based on the first 13 bits of the message, which represent the address field. The first 8 bits of the address represent the Node address and the other 5 bits represent the Sub-node address. These fields are used in a hierarchical addressing scheme where the first field identifies the bus nod and the second one selects the corresponding module from the device. The message routing is made based on a Routing Table which indicates the correspondence between the used addresses and the node ports.

The table content is defined by the initial configuration procedure of the interface.

*Rate i N MG M MG K MG e* \_ ( \_ 19 \_ 1) 1 3 (4.1)

and the last one is the Payload field on 128 bits.

Fig. 4.1 MF for 802.16 air interface standard for the 768Mbps line rate

parameters.

messages.

based on fixed MF structure.

*C. Transport Layer* 

The Application Layer builds the data messages. It maps data and control information into the message payload and attaches the message header. The payload is represented by concatenation of signal samples in the baseband. For the case of the 802.16 air interface, the format of a data message payload field is shown in Figure 4.2.

#### Fig. 4.2 802.16 payload mapping

All transfers over the bus are performed over paths. A path represents a connection between a source node and a destination node. The connection is made by a set of bus links defined by the routing tables. Physically, a path consists of a set of message slots per MF. Paths are defined before bus initialization and they remain fixed during operation, i.e. the transfer between two nodes is made on the same bus links using the same message slots. For each path, a message transmission rule is applied. There are two types of rules: mandatory low level rules, using modulo computation over message slot counters and optionally high level rules, using the Bit Map (BM) concept. In the second case each MG is formed as groups of messages from baseband channels. Each group may contain Ethernet frames as well.

#### **4.3 RP3-01 synchronization procedure**

In frame based communications systems, appropriate frequency/ time synchronization is critically important. In order to avoid inter-cell interference, all base stations must use the same frequency/ time reference. The transmission over the radio channel, both on downlink and uplink, should be synchronized with the same reference. The RP1 synchronization burst generator, located at Clock Control Manager (CCM) level, shall provide frame timing and time stamping for each of the air interface standard independently, based on that reference. The PR1 Frame Clock Bursts (FCB) is mapped into RP3 messages.

The synchronization algorithm uses several inputs and is realized based on information collected both at BBM and RRU. The first one is the propagation delay (PD) between BBM and RRU. This time interval is measured on BBM side using special message transmission called Round Trip Time (RTT) measurement message. The second input is the receiving time of the RP1 FCB. This interval is measured from the beginning of the last MF until the last bit of the FCB using a counter called *C1*. After measuring this time interval, the Frame

MicroTCA Compliant WiMAX BS Split Architecture with

MIMO Capabilities Support Based on OBSAI RP3-01 Interfaces 97

By selecting the right node/ sub-node address, the Application Layer from BBM OBSAI RP3-01 Interface selects the desired RRU. The Transport Layer from RRU1 OBSAI RP3-01 Interface directs the data streams to its own Application Layer when RRU1's address is used, otherwise forwards the data streams to the second Data Link Layer from the OBSAI RP3-01 Interface. In uplink (UL) direction (from RRU to BBM) the procedure is similar. Both receivers can be used (e. g. receive diversity or collaborative MIMO) or only one receiver can be active. In addition to these data and control streams that should be treated by the OBSAI RP3-01 Interface as RP3 streams, an Ethernet stream will be also transmitted between BBM and RRUs in order to connect the corresponding Control & Management (CM) modules to

RP3-01

RRU. This stream should be treated by the OBSAI RP3-01 Interface as RP1 stream.

Fig. 4.4 Proposed BS split architecture with interface protocol stack

Clock Synchronization (FCS) message is generated. FCS contains information from FCB and the *C1* value. The last input is the detection time of a FCS inside a MF. This time interval is measured by RRU using a counter called *C2*. Having all this information, RRU computes the buffering time (Brru). Figure 4.3 describes such an example. The time intervals are not at real scale. They are expressed in Time Units (TU). One can observe that at RRU side, the end of recovered FCB corresponds to the beginning of RF(k+4).

Using the formulas from [2], we obtain the Bruu as in (4.2), where *k* equals to 2 from some computing conditions.

$$Brru = k \cdot RFd + (C1 - PD) - FCBd = 11\,\text{TU} \tag{4.2}$$

Fig. 4.3 Timing principle in RP1 frame clock burst transfer

#### **4.4 Proposed implementation scheme**

The proposed BS split architecture is presented in Figure 4.4. A BBM is connected to the two RRUs in order to have multiple transmit/ receive antennas for MIMO capabilities. The connection between the two RRUs is realized using a chain topology. In order to obtain a single point failure redundancy scheme, a second BBM connected to the two RRUs is required. Only one BBM will be active at the time. Figure 4.4 depicts also the OBSAI RP3-01 Interfaces required for blocks interconnection. Note that on OBSAI RP3-01 interface of each RRU the same Transport and Application Layers serves the both Physical and Data Link Layers.

The RP3-01 connections between each BBM and RRU or between RRUs are bidirectional. On downlink (DL) direction (from BBM to RRU), the data stream from a BBM can contain multiple data streams interleaved/ multiplexed for the two Transceivers (e. g. in order to provide Space Time Coding or MIMO) or can contain data streams only for one Transceiver.

Clock Synchronization (FCS) message is generated. FCS contains information from FCB and the *C1* value. The last input is the detection time of a FCS inside a MF. This time interval is measured by RRU using a counter called *C2*. Having all this information, RRU computes the buffering time (Brru). Figure 4.3 describes such an example. The time intervals are not at real scale. They are expressed in Time Units (TU). One can observe that at RRU side, the end

Using the formulas from [2], we obtain the Bruu as in (4.2), where *k* equals to 2 from some

The proposed BS split architecture is presented in Figure 4.4. A BBM is connected to the two RRUs in order to have multiple transmit/ receive antennas for MIMO capabilities. The connection between the two RRUs is realized using a chain topology. In order to obtain a single point failure redundancy scheme, a second BBM connected to the two RRUs is required. Only one BBM will be active at the time. Figure 4.4 depicts also the OBSAI RP3-01 Interfaces required for blocks interconnection. Note that on OBSAI RP3-01 interface of each RRU the same Transport and Application Layers serves the both Physical and Data Link

The RP3-01 connections between each BBM and RRU or between RRUs are bidirectional. On downlink (DL) direction (from BBM to RRU), the data stream from a BBM can contain multiple data streams interleaved/ multiplexed for the two Transceivers (e. g. in order to provide Space Time Coding or MIMO) or can contain data streams only for one Transceiver.

*Brru k RFd C PD FCBd* ( 1 ) 11**TU** (4.2)

of recovered FCB corresponds to the beginning of RF(k+4).

Fig. 4.3 Timing principle in RP1 frame clock burst transfer

**4.4 Proposed implementation scheme** 

Layers.

computing conditions.

By selecting the right node/ sub-node address, the Application Layer from BBM OBSAI RP3-01 Interface selects the desired RRU. The Transport Layer from RRU1 OBSAI RP3-01 Interface directs the data streams to its own Application Layer when RRU1's address is used, otherwise forwards the data streams to the second Data Link Layer from the OBSAI RP3-01 Interface. In uplink (UL) direction (from RRU to BBM) the procedure is similar. Both receivers can be used (e. g. receive diversity or collaborative MIMO) or only one receiver can be active. In addition to these data and control streams that should be treated by the OBSAI RP3-01 Interface as RP3 streams, an Ethernet stream will be also transmitted between BBM and RRUs in order to connect the corresponding Control & Management (CM) modules to RRU. This stream should be treated by the OBSAI RP3-01 Interface as RP1 stream.

Fig. 4.4 Proposed BS split architecture with interface protocol stack

MicroTCA Compliant WiMAX BS Split Architecture with

Fig. 4.6 Block scheme for RRU1 RP3-01 interface

MIMO Capabilities Support Based on OBSAI RP3-01 Interfaces 99

Figure 4.5 presents the generation of RP3-01 stream at BBM side, while Figure 4.6 corresponds for RRU1 side. Each RP3-01 node has its own address known from the initial configuration procedure. The DL node addresses is represented with normal fonts and with bold italic fonts the UL node addresses. For the proposed system architecture, in DL direction the RP3-01 link is used to connect a source node with two destination nodes, so two paths exist over DL connection. Even if only one RRU transmitter is used at the time, the two paths will exist and the Transport Layer from BBM OBSAI RP3-01 Interface will place its messages over the message slots corresponding to selected RRU (consider twice the rate on link between BBM and RRU1 comparing with the one between RRU1 and RRU2). In the downlink direction, a point-to-point message transfer is applied, while in uplink direction, the same message is multicast to the two BBMs. Only the active BBM uses the received information.

Fig. 4.5 Block scheme for BBM RP3-01 interface

Each path will be considered having a data path and a control path. Bus manager will provide separate message transmission rules for the paths utilizing data and control message slots. We will explain the notations used in Figure 4.5 and Figure 4.6 and we will describe the steps made for RP3-01 interface generation.

Figure 4.5 presents the generation of RP3-01 stream at BBM side, while Figure 4.6 corresponds for RRU1 side. Each RP3-01 node has its own address known from the initial configuration procedure. The DL node addresses is represented with normal fonts and with bold italic fonts the UL node addresses. For the proposed system architecture, in DL direction the RP3-01 link is used to connect a source node with two destination nodes, so two paths exist over DL connection. Even if only one RRU transmitter is used at the time, the two paths will exist and the Transport Layer from BBM OBSAI RP3-01 Interface will place its messages over the message slots corresponding to selected RRU (consider twice the rate on link between BBM and RRU1 comparing with the one between RRU1 and RRU2). In the downlink direction, a point-to-point message transfer is applied, while in uplink direction, the same message is multicast to the two BBMs. Only the active BBM uses the

received information.

Fig. 4.5 Block scheme for BBM RP3-01 interface

describe the steps made for RP3-01 interface generation.

Each path will be considered having a data path and a control path. Bus manager will provide separate message transmission rules for the paths utilizing data and control message slots. We will explain the notations used in Figure 4.5 and Figure 4.6 and we will

Fig. 4.6 Block scheme for RRU1 RP3-01 interface

MicroTCA Compliant WiMAX BS Split Architecture with

Fig. 4.7 Interface Tx and Rx chains

**4.5 Conclusions** 

and receive antennas.

2006

[4] "CPRI Specification v4.0", June 2008

**5. References** 

MIMO Capabilities Support Based on OBSAI RP3-01 Interfaces 101

This section presented an overview of an OBSAI RP3-01 Interface implementation. It was described the main functions of the interface layers and there were presented interface block schemes for both BBM and RRU sides based on the protocol stack. Some examples were made for 802.16 air interface standard without reducing the generality of presentation. Base station split architecture was proposed, with support for redundancy and multiple transmit

[1] John C. Eidson, "Measurement, control and communication using IEEE 1588", Springer

[6] "PICMG Advanced Mezzanine Card Base Specification R 2.0", November 2006 [7] "Intelligent Platform Management Interface Specification – v2.0", February 2006 [8] "Trimble Resolution T system designer reference manual", www.trimble.com

[2] "OBSAI Reference Point 3 Specification", version 4.1, July 2008 [3] "OBSAI Reference Point 1 Specification", version 1.0, October 2003

[5] "PICMG MicroTCA Base Specification R1.0", July 2006

In DL direction, on BBM side, the Application Layer receives two data streams for the two RRUs, DT1 and DT2 and two control streams, CT1 and CT2. Also, an Ethernet stream for management called EthT and the RP1 FCBs are received. The Application Layer generates the corresponding messages steams, i.e. MDT1, MDT2, MCT1, MCT2, MEthT and RP1 FCS. Also specific RP3-01 link control messages are generated. This stream is called MLCT. Beside the message generator function, Application Layer is responsible for buffering the data paths. A buffer is required for each 802.16 signal (antenna-carrier) in order to compensate the jitter caused by message transmission. Finally, the Application Layer has to provide to Transport Layer the implementation specific message transmission rules. These rules could include the lower layer message transmission rules and/ or extra rules for mapping the RP1 traffic to RP3 data message slots.

The Transport Layer has four blocks with interleaving/ multiplexing function. First, the management messages, including RP1 FCS, MLCT and MEthT are put on the same flow called MCMT, based on a priority list. The two control messages streams MCT1 and MCT2 are multiplexed in the MCT flow. The data flow, called MDT is obtained by interleaving/ multiplexing the data messages from buffers and the management messages from MCMT stream. Finally, the Transport Layer multiplexes the MDT and the MCT streams based on the TDTI interface with the Transport Layer. Using this interface, the frame builder from Transport Layer requires data or control messages and increments the corresponding counters for each successful transfer. The generated RP3-01 frame, including also the special characters, is transferred to Physical Layer on PTTI interface. The DL continues on RRU side with the receiving chain form Figure 4.6. The Transport Layer works out the RP3-01 flow, first on data stream MDR and control stream MCR, and then the data path is split into data streams MDR1 and MDR2, respectively management messages RP1 FCS, MEthR and MLCR, while the control path is split into MCR1 stream and MCR2 stream. The Application Layer receives all these messages and extracts the corresponding payload. One interesting observation is that at RRU, the Transport Layer has also the message router function, as one can see from Figure 4.6.

In UL direction the procedure follows the same steps as the one described for DL.

For implementation we considered a XC4VFX60 device from Xilinx Virtex 5 family. The functional tests were made using ModelSIM 6.2g and the synthesis results were obtained using Xilinx ISE 9.2i. From the proposed architecture depicted in Figure 4.4 one can see that the RRU Interface contains two Data Link and two Physical Layers. The implementation results obtained for the Data Link Layer are critical for the global resources, while the Physical Layer implementation cost reflects in the number of used RocketIO Transceivers.

For this reason we start our implementation with the Data Link Layer. Figure 4.7 depicts the main blocks of Data Link Layer on the transmit chain, respective on the receive chain of the interface. The area and speed reports are presented in Table 4.1.


Table 4.1 Area and speed reports

Fig. 4.7 Interface Tx and Rx chains

#### **4.5 Conclusions**

100 Advanced Transmission Techniques in WiMAX

In DL direction, on BBM side, the Application Layer receives two data streams for the two RRUs, DT1 and DT2 and two control streams, CT1 and CT2. Also, an Ethernet stream for management called EthT and the RP1 FCBs are received. The Application Layer generates the corresponding messages steams, i.e. MDT1, MDT2, MCT1, MCT2, MEthT and RP1 FCS. Also specific RP3-01 link control messages are generated. This stream is called MLCT. Beside the message generator function, Application Layer is responsible for buffering the data paths. A buffer is required for each 802.16 signal (antenna-carrier) in order to compensate the jitter caused by message transmission. Finally, the Application Layer has to provide to Transport Layer the implementation specific message transmission rules. These rules could include the lower layer message transmission rules and/ or extra rules for

The Transport Layer has four blocks with interleaving/ multiplexing function. First, the management messages, including RP1 FCS, MLCT and MEthT are put on the same flow called MCMT, based on a priority list. The two control messages streams MCT1 and MCT2 are multiplexed in the MCT flow. The data flow, called MDT is obtained by interleaving/ multiplexing the data messages from buffers and the management messages from MCMT stream. Finally, the Transport Layer multiplexes the MDT and the MCT streams based on the TDTI interface with the Transport Layer. Using this interface, the frame builder from Transport Layer requires data or control messages and increments the corresponding counters for each successful transfer. The generated RP3-01 frame, including also the special characters, is transferred to Physical Layer on PTTI interface. The DL continues on RRU side with the receiving chain form Figure 4.6. The Transport Layer works out the RP3-01 flow, first on data stream MDR and control stream MCR, and then the data path is split into data streams MDR1 and MDR2, respectively management messages RP1 FCS, MEthR and MLCR, while the control path is split into MCR1 stream and MCR2 stream. The Application Layer receives all these messages and extracts the corresponding payload. One interesting observation is that at RRU, the Transport Layer has also the message router function, as one

In UL direction the procedure follows the same steps as the one described for DL.

interface. The area and speed reports are presented in Table 4.1.

For implementation we considered a XC4VFX60 device from Xilinx Virtex 5 family. The functional tests were made using ModelSIM 6.2g and the synthesis results were obtained using Xilinx ISE 9.2i. From the proposed architecture depicted in Figure 4.4 one can see that the RRU Interface contains two Data Link and two Physical Layers. The implementation results obtained for the Data Link Layer are critical for the global resources, while the Physical Layer implementation cost reflects in the number of used RocketIO Transceivers. For this reason we start our implementation with the Data Link Layer. Figure 4.7 depicts the main blocks of Data Link Layer on the transmit chain, respective on the receive chain of the

*Component No. of slices (from 25280) Speed (MHz)* 

FSM Sync Tx 8 407 Framer Tx 197 194 FSM Sync Rx 8 407 Framer Rx 274 185

mapping the RP1 traffic to RP3 data message slots.

can see from Figure 4.6.

Table 4.1 Area and speed reports

This section presented an overview of an OBSAI RP3-01 Interface implementation. It was described the main functions of the interface layers and there were presented interface block schemes for both BBM and RRU sides based on the protocol stack. Some examples were made for 802.16 air interface standard without reducing the generality of presentation. Base station split architecture was proposed, with support for redundancy and multiple transmit and receive antennas.

#### **5. References**


**6** 

*Spain* 

**Space-Time Adaptation and MIMO** 

*Centre Tecnològic de Telecomunicacions de Catalunya-CTTC, Parc Mediterrani de la Tecnologìa, Castelldefels, Barcelona,* 

The use of multiple antennas at the transmitter and/or receiver sides has been used for years in order to increase the signal to noise ratio at the receiver or for beam steering (a.k.a. beamforming) to reduce the amounts of interference at the receiver. However, one of the main benefits of using a Multiple Input Multiple Output (MIMO) channel is the increase of the channel capacity. Nowadays, by using different space-time-frequency coding techniques, orthogonal (or quasi orthogonal) virtual paths between transmitter and receiver can be obtained. These virtual paths can be used in order to increase the spectral efficiency or in order to increase the signal diversity (i.e. space diversity). In fact, a fundamental trade-off between diversity and multiplexing capabilities exists and must be considered when designing a multiple antenna system. In this chapter, the following

 We start by describing and reviewing the (ergodic) capacity of the MIMO channel in case of perfect channel state information at the transmitter (CSIT). Afterwards, the

 Then, we analyze different spatial adaptation and precoding mechanisms that can be applied to increase the performance of the system (either in terms of throughput or robustness). A new spatial adaptation algorithm proposed by the authors and called *Transmit Antenna and space-time Coding Selection* (TACS) is described showing some performance results that illustrate the improve on performance and/or

Finally, the well-known Space-Time coding techniques are reviewed, and a summary of

The present analysis is done following the general Linear Dispersion Codes framework, which is of special interest since it allows describing in an elegant way most of the space-

Before going into the details of how MIMO transmission can be carried out, it is important to have a look to the capacity of the MIMO channel. Usually, for capacity evaluation of

outage capacity is studied leading to the diversity-multiplexing trade-off.

the MIMO techniques adopted in WiMAX2 (IEEE 802.16e/m) is provided.

**1. Introduction** 

issues are addressed:

throughput.

time block codes existing in the literature.

**2. Characteristics of the MIMO channel** 

**Standardization Status** 

Ismael Gutiérrez and Faouzi Bader


### **Space-Time Adaptation and MIMO Standardization Status**

Ismael Gutiérrez and Faouzi Bader

*Centre Tecnològic de Telecomunicacions de Catalunya-CTTC, Parc Mediterrani de la Tecnologìa, Castelldefels, Barcelona, Spain* 

#### **1. Introduction**

102 Advanced Transmission Techniques in WiMAX

[9] L. Gasparini, O. Zadedyurina, G. Fontana, D. Macii, A. Boni, Y. Ofek, "A Digital Circuit

[10] Yu. S. Shmaliy, A. V. Marienko and A.V. Savchuk, "GPS-Based Optimal Kalman

http://www.xilinx.com/support/documentation/user\_guides/ug196.pdf

in Measurement Sardagna, Trento, Italy, 16-18 July 2007

[11] "PMC Sierra PM8358 QuadPHY 10GX" datasheet, March 2005 [12] "Virtex 5 FPGA RocketIO GTP Transceiver. User Guide",

and Time Interval Meeting

[13] IEEE Standard 802.16e, 2005

for Jitter Reduction of GPS-disciplined 1-pps Synchronization Signals", AMUEM 2007 – International Workshop on Advanced Methods for Uncertainty Estimation

Estimation of Time Error, Frequency Offset and Aging", 31st Annual Precise Time

The use of multiple antennas at the transmitter and/or receiver sides has been used for years in order to increase the signal to noise ratio at the receiver or for beam steering (a.k.a. beamforming) to reduce the amounts of interference at the receiver. However, one of the main benefits of using a Multiple Input Multiple Output (MIMO) channel is the increase of the channel capacity. Nowadays, by using different space-time-frequency coding techniques, orthogonal (or quasi orthogonal) virtual paths between transmitter and receiver can be obtained. These virtual paths can be used in order to increase the spectral efficiency or in order to increase the signal diversity (i.e. space diversity). In fact, a fundamental trade-off between diversity and multiplexing capabilities exists and must be considered when designing a multiple antenna system. In this chapter, the following issues are addressed:


The present analysis is done following the general Linear Dispersion Codes framework, which is of special interest since it allows describing in an elegant way most of the spacetime block codes existing in the literature.

#### **2. Characteristics of the MIMO channel**

Before going into the details of how MIMO transmission can be carried out, it is important to have a look to the capacity of the MIMO channel. Usually, for capacity evaluation of

Space-Time Adaptation and MIMO Standardization Status 105

*UPA k F*

 

where ║H║F2 stands for the Froebenius norm. Similar research studies have been undertaken regarding the effects of antenna correlation on the channel capacity. It is shown in [2] that for low antenna correlation values the optimum strategy is to allocate the same power to all the eigenmodes, whereas for high correlation values the optimum strategy is

When dealing with frequency/time-variant channels, one intrinsic characteristic of the channel is the diversity that can be achieved. In case of single input single output transmission, the general approach is to use coding and interleaving in the frequency and time domains so that one codeword is spread over the highest number possible of channel states. However, frequency and time diversity incur in a loss in bandwidth and/or transmission time delay. Alternatively, in case of multiple input multiple output channels the spatial dimension can be also exploited in order to increase the diversity without neither losing bandwidth nor increasing the transmission delay. Some metrics are defined to characterize the diversity. First, the *diversity gain* (or diversity order) is linked with the number of independent fading branches. Formally, it is defined as the negative asymptotic

) of the *log-log* plot of the average error probability *P* versus the average

log . log (6)

*P*

*<sup>d</sup>*

*g*

 *M M* **H H** 

2 2

 

2 2

, E log 1 E log 1 . (5)

*<sup>n</sup>*

allocating all the power to the strongest eigenmode.

Fig. 1. Array and diversity gain in Rayleigh fading channels.

**2.1 The diversity-multiplexing trade-off** 

slope (i.e. for

SNR  *C*

*k*

1

MIMO channels it is assumed that the fading coefficients between antenna pairs are i.i.d. Rayleigh distributed. In addition, and without loss of generality, it is also assumed that the channel is constant during the transmission of one MIMO codeword. Under this assumption the channel is referred as a block fading channel.

As shown in [1], the capacity of the MIMO channel can be obtained as follows

$$\mathbf{C}\left(\rho,\mathbf{H}\right) = \underset{\mathbf{Q}\succ 0:Tr\left[\mathbf{Q}\right]=M}{\arg\max} \log\_2 \det\left[\mathbf{I}\_N + \frac{\rho}{M}\mathbf{H}\mathbf{Q}\mathbf{H}^H\right] \tag{1}$$

where **H** is the MIMO channel matrix, *M* the number of transmit antennas, *IN* an identity matrix of size *N* equal to the number of receive antennas, is the Signal to Noise Ratio (SNR) and **Q** is the input covariance matrix whose trace is normalized to be equal to the number of transmit antennas. To gain a further insight on the channel characteristics, a good method is to apply the *Singular Value Decomposition* (SVD) to the MIMO channel matrix, so we express the channel matrix as

$$\mathbf{H} = \mathbf{U}\boldsymbol{\Sigma}\mathbf{V}^{\mathrm{H}},\tag{2}$$

where is a diagonal matrix, whose entries are the eigenvalues of **H**, and **U** and **V** are the lower and upper diagonal matrices respectively. The first important characteristic of the MIMO channel is given by the number of eigenvalues which tells us about the number of the independent virtual channels between transmitter and receiver. In addition, using the SVD of channel matrix, we can rewrite the channel capacity as

$$\mathbf{C}\left(\rho,\mathbf{H}\right) = \underset{\left\{\boldsymbol{p}\_{k}\right\}\_{k=1}^{n}}{\arg\max} \quad \sum\_{k=1}^{n} \log\_{2}\left(1 + \rho p\_{k}\sigma\_{k}^{2}\right) \tag{3}$$

where *<sup>k</sup>* and *pk* are the eigenvalues and the power transmitted through each of the said virtual channels respectively.

In case the channel is known at the transmitter, one can use this information to maximize the channel capacity by applying what it is known as multiple eigenmode transmission. This is much less complex that it sounds, and is carried out by multiplying the input vector **x** by **V**H. At the receiver side, similar operation is needed, therefore the result of multiplying the received signal by **U***H* is shown in the following equation

$$\mathbf{y} = \mathbf{U}\Sigma\mathbf{V}^{\mathrm{H}}\mathbf{V}\mathbf{x} + \mathbf{n} = \mathbf{U}\Sigma\mathbf{x} + \mathbf{n} \implies \hat{\mathbf{x}} = \mathbf{U}^{\mathrm{H}}\mathbf{U}\Sigma\mathbf{x} + \mathbf{U}^{\mathrm{H}}\mathbf{n} = \Sigma\mathbf{x} + \tilde{\mathbf{n}},\tag{4}$$

where it is observed that the number of eivenvalues determines the number of independent complex symbols that can be transmitted per each MIMO codeword. It can be also concluded that at low SNR values, the optimum allocation strategy will be to allocate all the available power to the strongest (or dominant) eigenmode, whereas at high SNRs the maximum capacity is obtained by allocating the same power to all the non-zero eigenmodes [1]. Actually, it is also proved that uniform power allocation (UPA) is the optimum strategy for fast fading channels where the transmitter is not able to capture the instantaneous channel state. The (ergodic) channel capacity in case of UPA allocation is given by

MIMO channels it is assumed that the fading coefficients between antenna pairs are i.i.d. Rayleigh distributed. In addition, and without loss of generality, it is also assumed that the channel is constant during the transmission of one MIMO codeword. Under this assumption

> *M* **Q Q <sup>H</sup> I HQH**

 H 2

where **H** is the MIMO channel matrix, *M* the number of transmit antennas, *IN* an identity

(SNR) and **Q** is the input covariance matrix whose trace is normalized to be equal to the number of transmit antennas. To gain a further insight on the channel characteristics, a good method is to apply the *Singular Value Decomposition* (SVD) to the MIMO channel matrix, so

where is a diagonal matrix, whose entries are the eigenvalues of **H**, and **U** and **V** are the lower and upper diagonal matrices respectively. The first important characteristic of the MIMO channel is given by the number of eigenvalues which tells us about the number of the independent virtual channels between transmitter and receiver. In addition, using the

*n*

In case the channel is known at the transmitter, one can use this information to maximize the channel capacity by applying what it is known as multiple eigenmode transmission. This is much less complex that it sounds, and is carried out by multiplying the input vector **x** by **V**H. At the receiver side, similar operation is needed, therefore the result of multiplying the

where it is observed that the number of eivenvalues determines the number of independent complex symbols that can be transmitted per each MIMO codeword. It can be also concluded that at low SNR values, the optimum allocation strategy will be to allocate all the available power to the strongest (or dominant) eigenmode, whereas at high SNRs the maximum capacity is obtained by allocating the same power to all the non-zero eigenmodes [1]. Actually, it is also proved that uniform power allocation (UPA) is the optimum strategy for fast fading channels where the transmitter is not able to capture the instantaneous

channel state. The (ergodic) channel capacity in case of UPA allocation is given by

2 1

**y U ΣV Vx n U Σx n x UUΣx Un Σx n** H H <sup>H</sup> ˆ , (4)

*<sup>k</sup>* and *pk* are the eigenvalues and the power transmitted through each of the said

*N*

**H U ΣV**<sup>H</sup> , (2)

is the Signal to Noise Ratio

, arg max log det (1)

 

*k k*

, arg max log 1 (3)

2

As shown in [1], the capacity of the MIMO channel can be obtained as follows

0:

*Tr M*

the channel is referred as a block fading channel.

*C*

we express the channel matrix as

virtual channels respectively.

where  matrix of size *N* equal to the number of receive antennas,

SVD of channel matrix, we can rewrite the channel capacity as

received signal by **U***H* is shown in the following equation

*n k k*

*p k C p* 

1

**<sup>H</sup>**

$$\mathbf{C}\_{\text{IIPA}}\left(\boldsymbol{\rho}, \mathbf{H}\right) = \mathbf{E}\left\{\sum\_{k=1}^{n} \log\_{2}\left(1 + \frac{\boldsymbol{\rho}}{M}\boldsymbol{\sigma}\_{k}^{2}\right)\right\} = \mathbf{E}\left\{\log\_{2}\left(1 + \frac{\boldsymbol{\rho}}{M} \left\|\mathbf{H}\right\|\_{\text{F}}^{2}\right)\right\}.\tag{5}$$

where ║H║F2 stands for the Froebenius norm. Similar research studies have been undertaken regarding the effects of antenna correlation on the channel capacity. It is shown in [2] that for low antenna correlation values the optimum strategy is to allocate the same power to all the eigenmodes, whereas for high correlation values the optimum strategy is allocating all the power to the strongest eigenmode.

Fig. 1. Array and diversity gain in Rayleigh fading channels.

#### **2.1 The diversity-multiplexing trade-off**

When dealing with frequency/time-variant channels, one intrinsic characteristic of the channel is the diversity that can be achieved. In case of single input single output transmission, the general approach is to use coding and interleaving in the frequency and time domains so that one codeword is spread over the highest number possible of channel states. However, frequency and time diversity incur in a loss in bandwidth and/or transmission time delay. Alternatively, in case of multiple input multiple output channels the spatial dimension can be also exploited in order to increase the diversity without neither losing bandwidth nor increasing the transmission delay. Some metrics are defined to characterize the diversity. First, the *diversity gain* (or diversity order) is linked with the number of independent fading branches. Formally, it is defined as the negative asymptotic slope (i.e. for ) of the *log-log* plot of the average error probability *P* versus the average SNR 

$$\mathbf{g}\_d = -\frac{\log\left(P\right)}{\log\left(\rho\right)}.\tag{6}$$

Space-Time Adaptation and MIMO Standardization Status 107

a function of the SNR, whereas the diversity gain gives us an idea on how fast the outage

The analysis of Eq. (11) at high SNR and uncorrelated Rayleigh channel leads to the diversity-multiplexing trade-off of the channel [3]. It has been shown that *gd* (*gs*,) is a piecewise linear function joining the (*gs*, *gd*(*gs*,)) points with *gs*={0,…, min(*N,M*)} and *gd*=(*Ngs*)(*M-gs*)*.* This trade-off is illustrated in the following Fig. 2. It is observed that maximum diversity is achieved when there is no spatial multiplexing gain (i.e. the transmission rate is fixed), whereas the maximum spatial multiplexing gain is achieved when the diversity gain

Fig. 2. Asymptotic diversity-multiplexing trade-off in uncorrelated Rayleigh channels.

complexity than STTC and for this reason they are usually preferred.

Based on the above introduction on MIMO channel characteristics, and the very important principle of spatial diversity versus spatial multiplexing tradeoff, we could now start studying the Space-Time MIMO encoding techniques. Analogously to channel coding in SISO links, two types of channel coding have been used for MIMO channels: block coding (referred as Space Time/Frequency Block Coding, STBC/SFBC) and convolutional coding (referred as Space Time Trellis Coding-STTC) [4][5]. For the STBC case, the codeword is only a function of the input bits, whereas the encoder output for the STTC is a function of the input bits and the encoder state. The inherent memory of the STTC provides an additional coding gain compared to the STBC at the expense of higher computational complexity [7][8]. However, since STBC transforms the MIMO channel into an equivalent scalar AWGN channel [6], the concatenation of traditional channel coding with STBC shows good performance and even outperforms STTC for low number of receive antennas (*M*,*N*2) [7] and the same number of encoder states. Furthermore, STBC/SFBC are of significantly less

probability decreases with the SNR.

is zero (the outage probability is kept fixed).

**2.2 Space-time coding over MIMO channels** 

Finally, the *coding gain* is defined as the SNR gain (observed as a left shift of the error curve). Then the coding gain *gc* is analytically expressed as

$$P\_e = \left(\frac{c}{\mathcal{g}\_c \mathcal{\rho}}\right)^{\alpha \mathcal{g}\_a} \tag{7}$$

where *Pe* is error probability, and and *c* are scaling constants depending on the modulation level, the coding scheme, and the channel characteristics. The *array gain ga* represents the decrease of average SNR due to coherent combining (*beamforming*) in case of multiple antennas at both transmitter or receiver sides, and it is formally expressed as

$$\mathcal{g}\_a = \frac{\rho\_{ma}}{\rho\_{sa}} \tag{8}$$

where *sa* is the average SNR for the SISO link, and *ma* is the average SNR for the MIMO link. The three different concepts are illustrated in Fig. 1 which shows the bit error rate of a QPSK transmission having an AWGN channel and an uncorrelated Rayleigh channel.

In case of multiple antennas at both sides of the link, multiple independent channels exist according to the rank of the channel matrix [1]. The multiplexing gain *gs* is defined as the ratio of the transmission rate *R*() to the capacity of an AWGN with array gain *ga* 

$$\log\_s = \frac{R\left(\rho\right)}{\log\_2\left(1 + \mathcal{g}\_a \rho\right)} \quad \underset{\rho \to \infty}{\Longrightarrow} R\left(\rho\right) = \mathcal{g}\_s \log\_2\left(\mathcal{g}\_a \rho\right) \tag{9}$$

where *R*() is the transmission rate.

For a slow fading channel (i.e. block fading channel), the maximum achievable rate for each codeword is a time variant quantity that depends on the instantaneous channel realizations. In this case, the outage probability *Pout* metric is preferred and is defined as the probability that a given channel realization cannot support a given rate *R* 

$$P\_{\rm out} \left( R \right) = \inf\_{\mathbf{Q} > 0: \mathcal{T}r(\mathbf{Q}) \le M} P \left( \log\_2 \det \left( \mathbf{I}\_N + \frac{\rho}{M} \mathbf{H} \mathbf{Q} \mathbf{H}^H \right) < R \right). \tag{10}$$

where *inf* stands for the Q that achieves the lower bound in terms of outage probability. Then, we can gain insights into the channel behaviour by analysing the outage probability as a function of the SNR and for a given transmission rate. Actually, we can establish a relationship between the diversity gain and the multiplexing gain via the outage probability *Pout* as

$$\log\_d(\mathcal{g}\_{s^\*}\rho) = -\frac{\partial \log\left(P\_{out}\left(R\right)\right)}{\partial \log\left(\rho\right)} \quad \underset{\rho \to \infty}{\Longrightarrow} P\_{out}\left(\rho\right) = \rho^{-\mathcal{g}\_d}.\tag{11}$$

Both the multiplexing gain and the diversity gain are upper bounded by *gsmin*(*N*,*M*) and *gdNM*. Intuitively, the multiplexing gain indicates the increase of the transmission rate as

Finally, the *coding gain* is defined as the SNR gain (observed as a left shift of the error curve).

*c <sup>c</sup> <sup>P</sup> g* 

modulation level, the coding scheme, and the channel characteristics. The *array gain ga* represents the decrease of average SNR due to coherent combining (*beamforming*) in case of multiple antennas at both transmitter or receiver sides, and it is formally expressed

> *ma <sup>a</sup> sa*

link. The three different concepts are illustrated in Fig. 1 which shows the bit error rate of a QPSK transmission having an AWGN channel and an uncorrelated Rayleigh channel.

In case of multiple antennas at both sides of the link, multiple independent channels exist according to the rank of the channel matrix [1]. The multiplexing gain *gs* is defined as the

*<sup>s</sup> s a <sup>a</sup>*

For a slow fading channel (i.e. block fading channel), the maximum achievable rate for each codeword is a time variant quantity that depends on the instantaneous channel realizations. In this case, the outage probability *Pout* metric is preferred and is defined as the probability

> *PR P <sup>R</sup> <sup>M</sup>* **Q Q**

where *inf* stands for the Q that achieves the lower bound in terms of outage probability. Then, we can gain insights into the channel behaviour by analysing the outage probability

relationship between the diversity gain and the multiplexing gain via the outage probability

*g Rg g g* 

) to the capacity of an AWGN with array gain *ga* 

*g* 

*e*

*a g*

, (7)

(8)

2

log log 1 (9)

**I HQH** H

*<sup>d</sup> out <sup>g</sup>*

inf log det . (10)

for a given transmission rate. Actually, we can establish a

 

log , . log (11)

*ma* is the average SNR for the MIMO

and *c* are scaling constants depending on the

Then the coding gain *gc* is analytically expressed as

*sa* is the average SNR for the SISO link, and

*R*

2

that a given channel realization cannot support a given rate *R* 

) is the transmission rate.

*out <sup>N</sup> Tr M*

*d s out P R g g P*

Both the multiplexing gain and the diversity gain are upper bounded by *gsmin*(*N*,*M*) and *gdNM*. Intuitively, the multiplexing gain indicates the increase of the transmission rate as

<sup>2</sup> 0:

where *Pe* is error probability, and

ratio of the transmission rate *R*(

as a function of the SNR and

as

where 

where *R*(

*Pout* as

a function of the SNR, whereas the diversity gain gives us an idea on how fast the outage probability decreases with the SNR.

The analysis of Eq. (11) at high SNR and uncorrelated Rayleigh channel leads to the diversity-multiplexing trade-off of the channel [3]. It has been shown that *gd* (*gs*,) is a piecewise linear function joining the (*gs*, *gd*(*gs*,)) points with *gs*={0,…, min(*N,M*)} and *gd*=(*Ngs*)(*M-gs*)*.* This trade-off is illustrated in the following Fig. 2. It is observed that maximum diversity is achieved when there is no spatial multiplexing gain (i.e. the transmission rate is fixed), whereas the maximum spatial multiplexing gain is achieved when the diversity gain is zero (the outage probability is kept fixed).

Fig. 2. Asymptotic diversity-multiplexing trade-off in uncorrelated Rayleigh channels.

#### **2.2 Space-time coding over MIMO channels**

Based on the above introduction on MIMO channel characteristics, and the very important principle of spatial diversity versus spatial multiplexing tradeoff, we could now start studying the Space-Time MIMO encoding techniques. Analogously to channel coding in SISO links, two types of channel coding have been used for MIMO channels: block coding (referred as Space Time/Frequency Block Coding, STBC/SFBC) and convolutional coding (referred as Space Time Trellis Coding-STTC) [4][5]. For the STBC case, the codeword is only a function of the input bits, whereas the encoder output for the STTC is a function of the input bits and the encoder state. The inherent memory of the STTC provides an additional coding gain compared to the STBC at the expense of higher computational complexity [7][8]. However, since STBC transforms the MIMO channel into an equivalent scalar AWGN channel [6], the concatenation of traditional channel coding with STBC shows good performance and even outperforms STTC for low number of receive antennas (*M*,*N*2) [7] and the same number of encoder states. Furthermore, STBC/SFBC are of significantly less complexity than STTC and for this reason they are usually preferred.

Space-Time Adaptation and MIMO Standardization Status 109

*Tr Tr MT*

If we impose some conditions on the set of basis matrices **A***q*, **B***q* with *q=*{*0,…,Q*-1} the mapping between the input symbols and the transmitted codeword **X** is unique and the symbols can be perfectly recovered. Substituting Eq. (15) into Eq. (13) and applying the *vec* operator on both sides of the expression, an equivalent real valued system equation can be

0 0

**n**

*y Hs n* (16)

1 1

**n**

**n**

*Q Q*

*M M*

1 1

*Q Q*

*s n*

. (15)

,

.

 

(17)

 

H H

 *<sup>Q</sup> q q qq*

**AA BB**

<sup>0</sup> 0 0

**y n H**

*Q Q <sup>Q</sup>*

where *n*  **(***0,* **½)** is the real vector noise with i.i.d. components. The equivalent real valued

00 00 10 10

*hh h h*

 

 

*Α Β Α Β*

*Α Β Α Β*

*q q n*

**3. Detection techniques: Linear vs non-linear schemes** 

 

 **AA BB h**

*n qq q q*

where **h***n* is the *n*-th row of the MIMO channel matrix **H**. Theoretically the maximum number of possible independent streams or channel modes of the effective MIMO channel

following the LDC design, the number of scalar modes that are excited is equal to the rank of HTH, which means equal to 2*Q* in the best case [18]. In addition, it is observed that it is possible to use a linear receiver only if *Q≤NT*, otherwise the system would be undetermined. When *Q<NT* the system is over-determined, and in consequence more reliability is given to each estimated symbol (i.e. spatial diversity is increased at the expense of spatial

Recovering the transmitted symbols within each codeword might become a challenging task depending on the set of basis matrices used. Moreover, we have already observed that the

*hh h h*

01 11 11 11 2 2

*N N QN QN NT Q*

2 2 2 1 , , *qq qq T M <sup>n</sup> <sup>M</sup>*

*2Q* (maximum number of singular values different than zero). However,

**AA B B <sup>h</sup>** *<sup>Α</sup> <sup>Β</sup> ;h* (18)

*q*

 

**y**

0

 

**y**

**y**

channel matrix **H** is obtained as

*Η*

 

.

matrix H is 2*NT*

multiplexing).

with:

*Q*

1

1

written as

0

1

#### **2.2.1 Space-time block coding system model**

In this section an example of a MIMO system with *M, N* transmitter and receiver antennas respectively communicating over a frequency flat-fading channel is assumed. A codeword **X** is transmitted over *T* channel accesses (symbols) over the *M* transmitting antennas, hence **X**=[**x**0 … **x***T*-1] with **x***i*C*<sup>M</sup><sup>1</sup>*. All the codewords are contained inside a codebook X, and each codeword contains the information from *Q* complex symbols. The ratio of symbols transmitted per codeword is defined as the spatial multiplexing rate *rs*=*Q*/*T*, where in case of *rs*=*M* the code is referred as *full-rate*. The transmission rate is given by *R*=*Qnb*/*T* [bits/s/Hz] where *nb* is the number of bits transmitted by each **x***i*(*j*) complex symbol. Moreover, the spreading of the symbols in the time and spatial domains leads to the increase of diversity, whereas by modifying *Q* we can modify the spatial multiplexing gain. In consequence, *M*×*N*×*T* determines the maximum diversity order, while *Q* defines the spatial multiplexing rate [10]. During any *ith* time instant (equivalent to a channel access), the transmitted and received signals are related as

$$\mathbf{y}\_{i} = \sqrt{\frac{\rho}{M}} \mathbf{H}\_{i} \mathbf{x}\_{i} + \mathbf{n}\_{i}, \quad i = \{0, \ldots, T - 1\}. \tag{12}$$

In Eq. (12) we have assumed the channel is constant within each channel Access. However, we can go one a step further and assume that the channel is constant during the transmission of one whole codeword transmission (*Tcoh*>>*T*) time, and in this case the channel dependency on the *ith* time instant subindex can be dropped such that **H***i*=**H** for any *i*={0,…,*T*-1}. Under these conditions the quasi-static block fading channel model can be assumed, and we can rewrite Eq. (12) as follows

$$\mathbf{Y}^T = \sqrt{\frac{\rho}{M}} \mathbf{H} \mathbf{X}^T + \mathbf{V}^T \Rightarrow \mathbf{Y} = \sqrt{\frac{\rho}{M}} \mathbf{X} \mathbf{H}^T + \mathbf{V} \tag{13}$$

where **X**C*TM* means the space-time transmitted codeword, **Y**C*T<sup>N</sup>* means the received space-time samples and **V**C*T<sup>N</sup>* represents the noise over each receive antenna during each channel access.

#### **2.2.2 Linear Dispersion Codes**

The Linear Dispersion Codes (LDC) class belongs to a subclass of the STBC codes where the codeword is given by a linear function of the input data symbols [11][20][21]. When the codeword is a linear function of the data symbols, the transmitted codeword can be expressed as

$$\mathbf{X} = \sum\_{q=1}^{Q} \left( \alpha\_q \mathbf{A}\_q + j \beta\_q \mathbf{B}\_q \right) \tag{14}$$

where **A***q* C*TM* determines how the real part of the symbol sq, *αq* , is spread over the spacetime domain, and the same for the imaginary part *βq* which is spread according to **B***q* C*TM* . For power normalization purposes, it is considered that the transmitted complex symbols *sq* have zero mean and unitary energy, this is E{*sq \*sq*}=1. The matrices **A***q*, and **B***q* are referred as the basis matrices and usually are normalized such that

$$\sum\_{q=0}^{Q-1} \left( Tr\left\{ \mathbf{A}\_q \mathbf{A}\_q^H \right\} + Tr\left\{ \mathbf{B}\_q \mathbf{B}\_q^H \right\} \right) = MT. \tag{15}$$

If we impose some conditions on the set of basis matrices **A***q*, **B***q* with *q=*{*0,…,Q*-1} the mapping between the input symbols and the transmitted codeword **X** is unique and the symbols can be perfectly recovered. Substituting Eq. (15) into Eq. (13) and applying the *vec* operator on both sides of the expression, an equivalent real valued system equation can be written as

$$\mathbf{y} = \begin{bmatrix} \Re(\mathbf{y}\_0) \\\\ \Im(\mathbf{y}\_0) \\\\ \vdots \\\\ \Re(\mathbf{y}\_{Q-1}) \\\\ \Im(\mathbf{y}\_{Q-1}) \end{bmatrix} = \sqrt{\frac{\rho}{M}} \mathbf{H} \begin{bmatrix} a\_0 \\\\ \beta\_0 \\\\ \vdots \\ a\_{Q-1} \\\\ \beta\_{Q-1} \end{bmatrix} + \begin{bmatrix} \mathbf{n}\_0 \\\\ \mathbf{n}\_0 \\\\ \vdots \\\\ \mathbf{n}\_{Q-1} \end{bmatrix} = \sqrt{\frac{\rho}{M}} \mathbf{H} \mathbf{\hat{s}} + \mathbf{n}\_{\prime} \tag{16}$$

where *n*  **(***0,* **½)** is the real vector noise with i.i.d. components. The equivalent real valued channel matrix **H** is obtained as

$$\mathcal{H} = \begin{bmatrix} \mathcal{A}\_0 \boldsymbol{\hbar}\_0 & \mathcal{B}\_0 \boldsymbol{\hbar}\_0 & \cdots & \mathcal{A}\_{\mathcal{Q}-1} \boldsymbol{\hbar}\_0 & \mathcal{B}\_{\mathcal{Q}-1} \boldsymbol{\hbar}\_0 \\ \vdots & \vdots & \ddots & \vdots & \vdots \\ \mathcal{A}\_0 \boldsymbol{\hbar}\_{N-1} & \mathcal{B}\_1 \boldsymbol{\hbar}\_{N-1} & \cdots & \mathcal{A}\_{\mathcal{Q}-1} \boldsymbol{\hbar}\_{N-1} & \mathcal{B}\_{\mathcal{Q}-1} \boldsymbol{\hbar}\_{N-1} \end{bmatrix}\_{2NT \times 2\mathcal{Q}}.\tag{17}$$

with:

108 Advanced Transmission Techniques in WiMAX

In this section an example of a MIMO system with *M, N* transmitter and receiver antennas respectively communicating over a frequency flat-fading channel is assumed. A codeword **X** is transmitted over *T* channel accesses (symbols) over the *M* transmitting antennas, hence

codeword contains the information from *Q* complex symbols. The ratio of symbols transmitted per codeword is defined as the spatial multiplexing rate *rs*=*Q*/*T*, where in case of *rs*=*M* the code is referred as *full-rate*. The transmission rate is given by *R*=*Qnb*/*T* [bits/s/Hz] where *nb* is the number of bits transmitted by each **x***i*(*j*) complex symbol. Moreover, the spreading of the symbols in the time and spatial domains leads to the increase of diversity, whereas by modifying *Q* we can modify the spatial multiplexing gain. In consequence, *M*×*N*×*T* determines the maximum diversity order, while *Q* defines the spatial multiplexing rate [10]. During any *ith* time instant (equivalent to a channel access),

*i ii i i T <sup>M</sup>* **y Hx n**

In Eq. (12) we have assumed the channel is constant within each channel Access. However, we can go one a step further and assume that the channel is constant during the transmission of one whole codeword transmission (*Tcoh*>>*T*) time, and in this case the channel dependency on the *ith* time instant subindex can be dropped such that **H***i*=**H** for any *i*={0,…,*T*-1}. Under these conditions the quasi-static block fading channel model can be

> *T TT T M M* **Y HX Y XH**

The Linear Dispersion Codes (LDC) class belongs to a subclass of the STBC codes where the codeword is given by a linear function of the input data symbols [11][20][21]. When the codeword is a linear function of the data symbols, the transmitted codeword can be

> *<sup>Q</sup> q q qq*

where **A***q* C*TM* determines how the real part of the symbol sq, *αq* , is spread over the spacetime domain, and the same for the imaginary part *βq* which is spread according to **B***q* C*TM* . For power normalization purposes, it is considered that the transmitted complex symbols *sq*

**X AB** 

*j*

 

*q*

1

*M* means the space-time transmitted codeword, **Y**C*T*

 

*<sup>1</sup>*. All the codewords are contained inside a codebook X, and each

, 0,..., 1 . (12)

*V V* (13)

*<sup>N</sup>* represents the noise over each receive antenna during each

, (14)

*\*sq*}=1. The matrices **A***q*, and **B***q* are referred as

*<sup>N</sup>* means the received

**2.2.1 Space-time block coding system model** 

the transmitted and received signals are related as

assumed, and we can rewrite Eq. (12) as follows

have zero mean and unitary energy, this is E{*sq*

the basis matrices and usually are normalized such that

**X**=[**x**0 … **x***T*-1] with **x***i*C*<sup>M</sup>*

where **X**C*T*

channel access.

expressed as

space-time samples and **V**C*T*

**2.2.2 Linear Dispersion Codes** 

$$\mathbf{J}\_{q} = \begin{bmatrix} \Re\left\{ \mathbf{A}\_{q} \right\} & -\Im\left\{ \mathbf{A}\_{q} \right\} \\ \Im\left\{ \mathbf{A}\_{q} \right\} & \Re\left\{ \mathbf{A}\_{q} \right\} \end{bmatrix}, \mathbf{B}\_{q} = \begin{bmatrix} -\Im\left\{ \mathbf{B}\_{q} \right\} & -\Re\left\{ \mathbf{B}\_{q} \right\} \\ \Re\left\{ \mathbf{B}\_{q} \right\} & -\Im\left\{ \mathbf{B}\_{q} \right\} \end{bmatrix} \in \mathbb{R}^{2T \times 2M}, \\ \mathbf{f}\_{n} = \begin{bmatrix} \Re\left\{ \mathbf{h}\_{n} \right\} \\ \Im\left\{ \mathbf{h}\_{n} \right\} \end{bmatrix} \in \mathbb{R}^{2M \times 1}, \tag{18}$$

where **h***n* is the *n*-th row of the MIMO channel matrix **H**. Theoretically the maximum number of possible independent streams or channel modes of the effective MIMO channel matrix H is 2*NT2Q* (maximum number of singular values different than zero). However, following the LDC design, the number of scalar modes that are excited is equal to the rank of HTH, which means equal to 2*Q* in the best case [18]. In addition, it is observed that it is possible to use a linear receiver only if *Q≤NT*, otherwise the system would be undetermined. When *Q<NT* the system is over-determined, and in consequence more reliability is given to each estimated symbol (i.e. spatial diversity is increased at the expense of spatial multiplexing).

#### **3. Detection techniques: Linear vs non-linear schemes**

Recovering the transmitted symbols within each codeword might become a challenging task depending on the set of basis matrices used. Moreover, we have already observed that the

Space-Time Adaptation and MIMO Standardization Status 111

inter-symbol interference (despite the noise vector might be increased due to the equalization) or to minimize the mean square error (i.e. the MMSE). The first design

> . *ZF M*

where **I***Q* is an identity matrix of size *Q*. Besides the lower computational cost of the linear receivers, another advantage of using them is that the channel effects can be perfectly estimated on a symbol basis, hence a closed expression for the *ESINR* per each transmitted

*diag*

where **D**=*diag*[**GH**], and I*self* =**GH-D** is the self-interference term. The full expression of ML

It has been already stated in previous sections that in case the transmitter has perfect channel information knowledge, the SNR at the receiver is maximized if the transmitter applies all the power over the dominant eigenmode of the channel. Moreover, in order to increase the throughput it may be preferable to transmit over all the non-zero eigenmodes of the channel allocating to each mode a power obtained following the water-filling algorithm [10]. However, both (dominant and multiple eigenmode transmission) beamforming techniques require that the transmitter knows perfectly the channel state information (CSI), and also that the channel doesn't change during a sufficiently large period to allow the CSI estimation and application of the beamforming. In consequence, the beamforming might be only applied for low mobility scenarios and where the channel can be accurately estimated.

On the other hand, according to the MMSE criterion the following equalizer is obtained

*MMSE Q*

*M*

*q*

*diag* 

*ESINR*

**4. Exploiting the transmit channel knowledge** 

†

*M* 

. <sup>1</sup> *H q*

*H H self self*

**GG I I**

**DD**

<sup>1</sup> <sup>1</sup> H. *<sup>H</sup>*

**G Η Η I Η** (22)

*q*

(23)

**G Η** (21)

criterion is known as the Zero-Forcing equalization where

symbol can be obtained as

and ZF receivers can be found in [14].

Fig. 3. Linear space-time precoding

ergodic channel is also a function of the basis matrices set. Therefore, a compromise between complexity and performance (in terms of spectral efficiency or decoding errors) exists which motivates the implementation and use of different LDC codes and decoding schemes.

#### **3.1 Maximum Likelihood (ML) decoding**

Optimum signal detection requires the maximization of the likelihood function over the discrete set of the code alphabet [13]. Mathematically, this can be expressed as

$$\hat{s} = \arg\min\_{\hat{s} \in \Sigma} \left\| \mathbf{y} - \sqrt{\frac{\rho}{M}} \mathbf{H} \mathbf{s} \right\|\_{F}^{2} \tag{19}$$

where is the space of all the transmitted symbol vectors with all the input data combinations having the same likelihood. Regarding the computational complexity of the ML detector, since each vector has a set of *2Q* symbols each one mapped over *log2*(*Z*) bits, the computational complexity is exponential with *Qlog2(Z)*.

The theoretical framework to understand the behavior of a MIMO system using an ML receiver has been extensively studied and analyzed in the scientific literature (e.g. [9]-[13]) leading to important conclusions. The first and more obvious conclusion is that the diversity order is *g*d*N·min*(*M,T*) [13]. So, it becomes clear that *T* should be equal to *M* to achieve full diversity. However, increasing *T* requires increasing also the value of *Q* to maintain the same rate *R*, which leads to an increase in memory requirements, computational complexity, and delay. It is then apparent that a trade-off exists between achievable diversity and the decoding complexity.

Often the performance of a system is measured in terms of the post-processing SNR or the Effective SNR (ESINR). This ESINR value estimates the SNR required in an AWGN channel to obtain the same performance as in the given system (i.e. our MIMO system). In [14], the author proposed a simple parametrizable expression to estimate the performance of the system under different MIMO transmission schemes and antenna configurations. This model has been used in later sections when the performance evaluation of adaptive MIMO systems with ML receivers is developed and analysed.

#### **3.2 Linear detectors: Zero forcing and minimum mean square error**

The high computational cost of the ML receiver (O(2*Qlog2(Z)*)) makes the use of less computational demanding receiving techniques more appealing, sometimes even despite a degradation on the system performances. Following the expression in Eq. (12), a linear relationship between input and output symbols exists and the system can be solved applying simple algebra as long as *QNT*, i.e.

$$
\hat{\mathbf{s}} = \mathbf{G}\mathbf{y} = \sqrt{\frac{\rho}{M}} \mathbf{G} \mathbf{H} \mathbf{s} + \mathbf{G} \mathbf{n},\tag{20}
$$

where *G 2Q2N* is the equalizer matrix which compensates the MIMO channel effects. Similar to frequency equalization, the equalizer matrix might be designed to suppress the

ergodic channel is also a function of the basis matrices set. Therefore, a compromise between complexity and performance (in terms of spectral efficiency or decoding errors) exists which motivates the implementation and use of different LDC codes and decoding schemes.

Optimum signal detection requires the maximization of the likelihood function over the

where is the space of all the transmitted symbol vectors with all the input data combinations having the same likelihood. Regarding the computational complexity of the ML detector, since each vector has a set of *2Q* symbols each one mapped over *log2*(*Z*) bits,

The theoretical framework to understand the behavior of a MIMO system using an ML receiver has been extensively studied and analyzed in the scientific literature (e.g. [9]-[13]) leading to important conclusions. The first and more obvious conclusion is that the diversity order is *g*d*N·min*(*M,T*) [13]. So, it becomes clear that *T* should be equal to *M* to achieve full diversity. However, increasing *T* requires increasing also the value of *Q* to maintain the same rate *R*, which leads to an increase in memory requirements, computational complexity, and delay. It is then apparent that a trade-off exists between achievable diversity and the

Often the performance of a system is measured in terms of the post-processing SNR or the Effective SNR (ESINR). This ESINR value estimates the SNR required in an AWGN channel to obtain the same performance as in the given system (i.e. our MIMO system). In [14], the author proposed a simple parametrizable expression to estimate the performance of the system under different MIMO transmission schemes and antenna configurations. This model has been used in later sections when the performance evaluation of adaptive MIMO

The high computational cost of the ML receiver (O(2*Qlog2(Z)*)) makes the use of less computational demanding receiving techniques more appealing, sometimes even despite a degradation on the system performances. Following the expression in Eq. (12), a linear relationship between input and output symbols exists and the system can be solved

> <sup>ˆ</sup> , *<sup>M</sup>*

where *G 2Q2N* is the equalizer matrix which compensates the MIMO channel effects. Similar to frequency equalization, the equalizer matrix might be designed to suppress the

**s Gy GHs Gn** (20)

2

**Η** (19)

discrete set of the code alphabet [13]. Mathematically, this can be expressed as

the computational complexity is exponential with *Qlog2(Z)*.

systems with ML receivers is developed and analysed.

applying simple algebra as long as *Q*

**3.2 Linear detectors: Zero forcing and minimum mean square error** 

*NT*, i.e.

ˆ <sup>ˆ</sup> arg min *<sup>s</sup> <sup>F</sup> s y s M*

**3.1 Maximum Likelihood (ML) decoding** 

decoding complexity.

inter-symbol interference (despite the noise vector might be increased due to the equalization) or to minimize the mean square error (i.e. the MMSE). The first design criterion is known as the Zero-Forcing equalization where

$$\mathbf{G}\_{ZF} = \sqrt{\frac{M}{\rho}} \mathbf{H}^{\dagger}. \tag{21}$$

On the other hand, according to the MMSE criterion the following equalizer is obtained

$$\mathbf{G}\_{MMSE} = \sqrt{\frac{M}{\rho}} \left( \mathbf{H}^H \mathbf{H} + \left( \frac{\rho}{M} \right)^{-1} \mathbf{I}\_Q \right)^{-1} \mathbf{H}^H. \tag{22}$$

where **I***Q* is an identity matrix of size *Q*. Besides the lower computational cost of the linear receivers, another advantage of using them is that the channel effects can be perfectly estimated on a symbol basis, hence a closed expression for the *ESINR* per each transmitted symbol can be obtained as

$$ESINR\_q = \frac{diag\left[\mathbf{D}\mathbf{D}^H\right]\_q}{diag\left[\frac{1}{\rho}\mathbf{G}^H\mathbf{G} + \mathbf{I}\_{\text{self}}\mathbf{I}\_{\text{self}}^H\right]\_q} \tag{23}$$

where **D**=*diag*[**GH**], and I*self* =**GH-D** is the self-interference term. The full expression of ML and ZF receivers can be found in [14].

#### **4. Exploiting the transmit channel knowledge**

It has been already stated in previous sections that in case the transmitter has perfect channel information knowledge, the SNR at the receiver is maximized if the transmitter applies all the power over the dominant eigenmode of the channel. Moreover, in order to increase the throughput it may be preferable to transmit over all the non-zero eigenmodes of the channel allocating to each mode a power obtained following the water-filling algorithm [10]. However, both (dominant and multiple eigenmode transmission) beamforming techniques require that the transmitter knows perfectly the channel state information (CSI), and also that the channel doesn't change during a sufficiently large period to allow the CSI estimation and application of the beamforming. In consequence, the beamforming might be only applied for low mobility scenarios and where the channel can be accurately estimated.

Fig. 3. Linear space-time precoding

Space-Time Adaptation and MIMO Standardization Status 113

As a result, we can note that RAS implies a loss in the SNR which becomes larger as the difference between *N* and *Na* is increased. However, the diversity order for both schemes is exactly the same and there is only a coding gain difference [19]. The analysis for Transmit Antenna Selection (TAS) is reciprocal, therefore, the same effects are observed in case of

For the MIMO case and if multiple streams are simultaneously transmitted, transmit and/or receive antenna selection has further implications than just a reduction of the array gain. Actually, the inherent spatial multiplexing-diversity trade-off leads to different optimization criteria: diversity optimization (i.e. select the set of antennas that gives a higher Frobenius norm of the channel), improve the link reliability (discard antennas that produces large fadings in any eigenmode), maximize the Shannon capacity, etc. Furthermore, in [19] it is stated that the diversity gain obtained by transmit antenna and receive antenna selection is the same as without selection procedure, hence *gd*=(*N-gs*)(*M-gs*) with *gs*={0,…, min

Transmit antenna selection techniques were first proposed during the very late 90s in the context of MIMO links in order to improve the array gain. During that period, antenna selection was derived according to the class of ST coding scheme that was involved. Heath et al. in [22] focused on the antenna selection in case of Spatial Multiplexing for linear receivers. The optimization criterion in [22] was to maximize the post-processing SNR in order to minimize the bit error probability. Later, it has been shown that the difference between optimizing the *ESINR* is only 0.5dB better than optimizing the lowest eigenvalue (the ESINR in case of Zero-Forcing is lower bounded by the minimum eigenvalue of H). Similar works have been carried out for Orthogonal STBC (OSTBC) which in this case concluded that maximizing the Frobenius norm of the active channel was the optimum strategy [23]. More recently, Deng et al. extended these transmit antenna selection schemes under the LDC framework concluding that the best selection criteria for minimize the bit error probability is based on maximizing the post-processing (or Effective) SNR [24]. Finally, an interesting application of transmit antenna selection has been proposed by Freitas et al. in [25] where different spatial layers are assumed combining spatial diversity and spatial multiplexing. In [25] the different branches are disposed in parallel hence both spatial diversity and multiplexing gains can be simultaneously achieved. The antennas subsets are then assigned to the spatial layers in order to minimize the bit error probability, where the (more susceptible) SM based layers are assigned the best subset of antennas and the

In the previous section, TAS precoding scheme has been introduced for some of the existing STBC and LDC. Therefore, given a specific code, the number of bits *fb* that must be fed-back

*<sup>a</sup> <sup>a</sup> <sup>b</sup>*

*<sup>M</sup> <sup>M</sup> <sup>f</sup> <sup>M</sup> MM M* 

! . ! ! (25)

*a*

MISO with TAS.

(*N,M*)}.

**4.2 Transmit antenna selection in MIMO systems** 

remaining are assigned to the OSTBC layers.

**4.3 MIMO precoding based on LDC codes** 

from the receiver to indicate the optimum transmit antennas set is

In order to obtain the channel information at the transmitter, the most common approach is that the receiver sends some signalling to allow the transmitter to know the status of the downlink channel (in case of FDD system this has to be done explicitly by transmitting the matrix **H**, and for TDD systems the channel reciprocity allows sending some pilots in the reverse link so that the channel is estimated for the forward channel). However, any of these two alternatives will consume bandwidth either in the form of feedback signalling or channel estimation signalling. This triggered lot of work on how to reduce the feedback leading to techniques and metrics such as quantizing the channel information, using the channel condition number, the Demmel condition number, the channel rank, etc [26]. In general, the schemes where the input symbols are adjusted according to the channel status are known as precoders. Actually, precoding and space-time block coding can be considered into the same block where given a set of codes (defined by the codebook) one of them is selected each *Time Transmission Interval* (defined by *T* or multiples of) according to the *bf* feedback bits.

#### **4.1 Transmit and receive antenna selection**

Besides the increase of the capacity or the reliability by any of the before mentioned precoding techniques (beamforming, codeword selection based on finite codebooks, etc.), a very simple precoding technique is to select which antenna (or subset of antennas) should be used according to an optimization criterion (e.g. capacity, reliability, etc.). Antenna selection also aids to reduce the hardware cost as well as the signal processing requirements, therefore, it may be good for handheld receivers where, space, power consumption, and cost must be seriously taken into account. Obviously, the reduction of the number of antennas reduces the array gain, however when the channel in any of these antennas is experiencing a deep fade, the capacity loss by not using such antenna is negligible [19]. In consequence, antenna selection at both transmitter and receiver helps in reducing the implementation costs while retaining most of the benefits of MIMO technology.

A MIMO system model considering antenna selection is depicted in Fig. 4, where *M* and *N* are the number of transmitter and receiver RF chains respectively, whereas the available antennas are referred by *Ma* and *Na* for transmitter and receiver respectively (*MMa , NNa*).

Fig. 4. Antenna selection in MIMO systems with *Ma* available transmit and *Na* receive antennas.

In the SIMO case, it is shown in [10] and [19] that the array gain using a Maximum Ratio Combiner (MRC) without Receive Antenna Selection (RAS) is equal to *ga=Na*, whereas when RAS is applied (e.g. the antenna with better channel is selected) the array gain is given by

$$\mathbf{g}\_a = \mathbf{N}\left(\mathbf{1} + \sum\_{j=N+1}^{N\_a} \frac{\mathbf{1}}{j}\right). \tag{24}$$

In order to obtain the channel information at the transmitter, the most common approach is that the receiver sends some signalling to allow the transmitter to know the status of the downlink channel (in case of FDD system this has to be done explicitly by transmitting the matrix **H**, and for TDD systems the channel reciprocity allows sending some pilots in the reverse link so that the channel is estimated for the forward channel). However, any of these two alternatives will consume bandwidth either in the form of feedback signalling or channel estimation signalling. This triggered lot of work on how to reduce the feedback leading to techniques and metrics such as quantizing the channel information, using the channel condition number, the Demmel condition number, the channel rank, etc [26]. In general, the schemes where the input symbols are adjusted according to the channel status are known as precoders. Actually, precoding and space-time block coding can be considered into the same block where given a set of codes (defined by the codebook) one of them is selected each *Time* 

*Transmission Interval* (defined by *T* or multiples of) according to the *bf* feedback bits.

Besides the increase of the capacity or the reliability by any of the before mentioned precoding techniques (beamforming, codeword selection based on finite codebooks, etc.), a very simple precoding technique is to select which antenna (or subset of antennas) should be used according to an optimization criterion (e.g. capacity, reliability, etc.). Antenna selection also aids to reduce the hardware cost as well as the signal processing requirements, therefore, it may be good for handheld receivers where, space, power consumption, and cost must be seriously taken into account. Obviously, the reduction of the number of antennas reduces the array gain, however when the channel in any of these antennas is experiencing a deep fade, the capacity loss by not using such antenna is negligible [19]. In consequence, antenna selection at both transmitter and receiver helps in reducing the implementation

A MIMO system model considering antenna selection is depicted in Fig. 4, where *M* and *N* are the number of transmitter and receiver RF chains respectively, whereas the available antennas are referred by *Ma* and *Na* for transmitter and receiver respectively (*MMa , NNa*).

Fig. 4. Antenna selection in MIMO systems with *Ma* available transmit and *Na* receive

*a*

In the SIMO case, it is shown in [10] and [19] that the array gain using a Maximum Ratio Combiner (MRC) without Receive Antenna Selection (RAS) is equal to *ga=Na*, whereas when RAS is applied (e.g. the antenna with better channel is selected) the array gain is given by

*Na*

<sup>1</sup> 1 . (24)

 1

*j N g N <sup>j</sup>*

**4.1 Transmit and receive antenna selection** 

antennas.

costs while retaining most of the benefits of MIMO technology.

As a result, we can note that RAS implies a loss in the SNR which becomes larger as the difference between *N* and *Na* is increased. However, the diversity order for both schemes is exactly the same and there is only a coding gain difference [19]. The analysis for Transmit Antenna Selection (TAS) is reciprocal, therefore, the same effects are observed in case of MISO with TAS.

For the MIMO case and if multiple streams are simultaneously transmitted, transmit and/or receive antenna selection has further implications than just a reduction of the array gain. Actually, the inherent spatial multiplexing-diversity trade-off leads to different optimization criteria: diversity optimization (i.e. select the set of antennas that gives a higher Frobenius norm of the channel), improve the link reliability (discard antennas that produces large fadings in any eigenmode), maximize the Shannon capacity, etc. Furthermore, in [19] it is stated that the diversity gain obtained by transmit antenna and receive antenna selection is the same as without selection procedure, hence *gd*=(*N-gs*)(*M-gs*) with *gs*={0,…, min (*N,M*)}.

#### **4.2 Transmit antenna selection in MIMO systems**

Transmit antenna selection techniques were first proposed during the very late 90s in the context of MIMO links in order to improve the array gain. During that period, antenna selection was derived according to the class of ST coding scheme that was involved. Heath et al. in [22] focused on the antenna selection in case of Spatial Multiplexing for linear receivers. The optimization criterion in [22] was to maximize the post-processing SNR in order to minimize the bit error probability. Later, it has been shown that the difference between optimizing the *ESINR* is only 0.5dB better than optimizing the lowest eigenvalue (the ESINR in case of Zero-Forcing is lower bounded by the minimum eigenvalue of H). Similar works have been carried out for Orthogonal STBC (OSTBC) which in this case concluded that maximizing the Frobenius norm of the active channel was the optimum strategy [23]. More recently, Deng et al. extended these transmit antenna selection schemes under the LDC framework concluding that the best selection criteria for minimize the bit error probability is based on maximizing the post-processing (or Effective) SNR [24]. Finally, an interesting application of transmit antenna selection has been proposed by Freitas et al. in [25] where different spatial layers are assumed combining spatial diversity and spatial multiplexing. In [25] the different branches are disposed in parallel hence both spatial diversity and multiplexing gains can be simultaneously achieved. The antennas subsets are then assigned to the spatial layers in order to minimize the bit error probability, where the (more susceptible) SM based layers are assigned the best subset of antennas and the remaining are assigned to the OSTBC layers.

#### **4.3 MIMO precoding based on LDC codes**

In the previous section, TAS precoding scheme has been introduced for some of the existing STBC and LDC. Therefore, given a specific code, the number of bits *fb* that must be fed-back from the receiver to indicate the optimum transmit antennas set is

$$\,\_{a}f\_{b} = \binom{M\_{a}}{M} = \frac{M\_{a}!}{M!(M\_{a}-M)!}.\tag{25}$$

Space-Time Adaptation and MIMO Standardization Status 115

the transmit antenna subset and the LDC code that minimizes the error rate probability (i.e. the bit error rate – BER) while the modulation that is required by each LDC is adapted in order to achieve the required rate *R*. In that case, since the *Q*-function is monotonically

max min , , , (26)

*ESINR H LDC p d Z* <sup>2</sup>

decreasing as a function of the input, the optimization problem is defined as follows

*q ii i LDC p <sup>q</sup>*

min ,

where *i* means the LDC index, *pi* denotes the transmitting antenna subset, *q* refers to the spatial stream (i.e. the symbol) index, and *dmin* is the minimum Euclidean distance according to the QAM constellation size used (note that the QAM constellation is a function of the

In the second scenario, the optimization is performed in order to maximize the system throughput considering a certain quality of service requirement (i.e. a maximum Block Error

where *j* means the Modulation and Coding Scheme (MCS) index that maximizes the spectral efficiency for the specific channel state subject to a maximum Block Error Rate (BLER). The following Fig. 5 illustrates the scheme of the MIMO system applying the TACS selection

Fig. 5. Proposed TACS spatial adaptation scheme and integration into the transmission

In this section we are setting the main parameters to evaluate the performance of the TACS scheme, the IEEE 802.16 standard is here used to carry out the experiment. Some parameters are depicted in Table 1, where perfect synchronization is assumed and inter-cell interference is not considered. The used modulation is a Z-QAM (Z={2,4,16,64}) with Gray mapping. According to the CSI measured, the BS determines: *i*) the antenna subset, *ii*) the LDC subset and in case of throughput maximization, *iii*) the MCS that maximizes the rate for a maximum Block Error Rate - BLER (second optimization criterion). The codebook is composed mainly by the Single Input Multiple Output (Maximum Ratio Combining is used at the receiver) receiver, the Alamouti 's Spatial Diversity (SD) coding scheme (referred as G2 in hereafter plotted the figures) [15], the pure spatial multiplexing (SM) and the Golden code [17]. The performance of the system is evaluated over 100.000 channel realizations,

*R BLER ESINR BLER*

max min 1 s.t.: (27)

<sup>210</sup> ... , , *b bb*

*i i*

Rate - BLER). In that case, the problem is formulated as follows

*<sup>q</sup> LDC p MCS <sup>q</sup>*

*ii j*

, ,

**4.4.2 TACS performance evaluation** 

LDC).

algorithm.

scheme.

However, if we could afford sending few more bits over the feedback channel, the transmitter/receiver may be able to select which code is more suitable according to the current channel state, or to choose how many spatial streams can be transmitted according to the channel rank [26]. Recent researches have extended the space-time coding selection (i.e. codebook based precoding) into the LDC framework [27]-[32]. An important result was obtained in [29], where it is shown that *fb*=*log2(M)* feedback bits are enough to achieve full diversity. In [32], the authors showed that the average SNR can be improved up to 2dB compared to the open loop scheme with only 3 feedback bits (i.e. 8 sets of LDC codes).

#### **4.4 Transmit antenna and space-time Code Selection (TACS)**

As it has been explained in the previous section, when partial CSIs information is available at the transmitter two common selection techniques could be applied, which are: the spacetime code selection, and the transmit antenna selection. One of the first works joining both concepts is that presented by Heath et al. in [33] where the number of the spatial streams (in the SM case) are adapted by selecting the best set of transmitter antennas (i.e. *fb*=*M*). Furthermore, it was stated that if the optimum number of streams are transmitted from the optimum selected antenna set, the diversity gain is also maximized (*gd≤ MN*). Then, given an antenna subset and a fixed rate, the required constellation could be determined as well as the number of spatial streams.

A simplification of this optimization problem is given in [34] where each stream is switched on/off when the post-processing of the SNR value of the stream is above/below a fixed threshold which is related with the rate. Further extensions of space-time code selection with TAS are given by Machado et al. in [35] where the available codes in the codebook are; the Alamouti code, the SM with *M*=2, the Quasi-OSTBC with *M*=3 and single antenna transmission.

In addition, the space-time code selection with transmit antenna selection has been generalized by the authors in [36]-[39] under the LDC framework considering both the linear and the ML receivers and developed within the IEEE 802.16m framework [38]. This generalization allows us to use any type of linear STBC (independently of the optimization criteria) codes and determine which codes are used most of the time and under which channel conditions. Two optimizations criteria have been developed in [14], one following the classical bit error rate optimization (minimizing of the scaled minimum Euclidian distance), and a second one is based on the throughput maximization given a fixed link quality (i.e. fixed packet error rate or bit error rate). This second optimization criterion can be used for resource allocation and scheduling purposes. Nevertheless, it is also shown that for low multiplexing rates the classic STBC codes (i.e. Alamouti, SM and Golden code) with transmit antenna selection are sufficient to explore the Grassmanian subspace [14].

#### **4.4.1 The TACS selection criteria**

Given the *ESINR* per stream and the average pairwise error probabilities, two different code and antenna subset optimization scenarios namely *Minimizing the bit error rate* and *Maximizing the throughput* respectively have been evaluated in [14][36]-[39]. In the first scenario, we consider that the same modulation is applied to all the symbols with a fixed rate *R*. In that case, and since transmission power is also fixed, we are interested in selecting

However, if we could afford sending few more bits over the feedback channel, the transmitter/receiver may be able to select which code is more suitable according to the current channel state, or to choose how many spatial streams can be transmitted according to the channel rank [26]. Recent researches have extended the space-time coding selection (i.e. codebook based precoding) into the LDC framework [27]-[32]. An important result was obtained in [29], where it is shown that *fb*=*log2(M)* feedback bits are enough to achieve full diversity. In [32], the authors showed that the average SNR can be improved up to 2dB compared to the open loop scheme with only 3 feedback bits (i.e. 8 sets of LDC codes).

As it has been explained in the previous section, when partial CSIs information is available at the transmitter two common selection techniques could be applied, which are: the spacetime code selection, and the transmit antenna selection. One of the first works joining both concepts is that presented by Heath et al. in [33] where the number of the spatial streams (in the SM case) are adapted by selecting the best set of transmitter antennas (i.e. *fb*=*M*). Furthermore, it was stated that if the optimum number of streams are transmitted from the optimum selected antenna set, the diversity gain is also maximized (*gd≤ MN*). Then, given an antenna subset and a fixed rate, the required constellation could be determined as well as

A simplification of this optimization problem is given in [34] where each stream is switched on/off when the post-processing of the SNR value of the stream is above/below a fixed threshold which is related with the rate. Further extensions of space-time code selection with TAS are given by Machado et al. in [35] where the available codes in the codebook are; the Alamouti code, the SM with *M*=2, the Quasi-OSTBC with *M*=3 and single antenna

In addition, the space-time code selection with transmit antenna selection has been generalized by the authors in [36]-[39] under the LDC framework considering both the linear and the ML receivers and developed within the IEEE 802.16m framework [38]. This generalization allows us to use any type of linear STBC (independently of the optimization criteria) codes and determine which codes are used most of the time and under which channel conditions. Two optimizations criteria have been developed in [14], one following the classical bit error rate optimization (minimizing of the scaled minimum Euclidian distance), and a second one is based on the throughput maximization given a fixed link quality (i.e. fixed packet error rate or bit error rate). This second optimization criterion can be used for resource allocation and scheduling purposes. Nevertheless, it is also shown that for low multiplexing rates the classic STBC codes (i.e. Alamouti, SM and Golden code) with

transmit antenna selection are sufficient to explore the Grassmanian subspace [14].

Given the *ESINR* per stream and the average pairwise error probabilities, two different code and antenna subset optimization scenarios namely *Minimizing the bit error rate* and *Maximizing the throughput* respectively have been evaluated in [14][36]-[39]. In the first scenario, we consider that the same modulation is applied to all the symbols with a fixed rate *R*. In that case, and since transmission power is also fixed, we are interested in selecting

**4.4 Transmit antenna and space-time Code Selection (TACS)** 

the number of spatial streams.

**4.4.1 The TACS selection criteria** 

transmission.

the transmit antenna subset and the LDC code that minimizes the error rate probability (i.e. the bit error rate – BER) while the modulation that is required by each LDC is adapted in order to achieve the required rate *R*. In that case, since the *Q*-function is monotonically decreasing as a function of the input, the optimization problem is defined as follows

$$\max\_{LDC\_i, p\_i} \min\_q \left\{ ESINR\_q\left(H, LDC\_i, p\_i\right) d\_{\min}^2\left(Z\_i\right) \right\},\tag{26}$$

where *i* means the LDC index, *pi* denotes the transmitting antenna subset, *q* refers to the spatial stream (i.e. the symbol) index, and *dmin* is the minimum Euclidean distance according to the QAM constellation size used (note that the QAM constellation is a function of the LDC).

In the second scenario, the optimization is performed in order to maximize the system throughput considering a certain quality of service requirement (i.e. a maximum Block Error Rate - BLER). In that case, the problem is formulated as follows

$$\max\_{LLR\_i, p\_i, MCS\_j} \min\_q R \left( 1 - BLRR \left( ESINR\_q \right) \right) \quad \text{s.t.: } BLRR \le \mu \tag{27}$$

where *j* means the Modulation and Coding Scheme (MCS) index that maximizes the spectral efficiency for the specific channel state subject to a maximum Block Error Rate (BLER). The following Fig. 5 illustrates the scheme of the MIMO system applying the TACS selection algorithm.

Fig. 5. Proposed TACS spatial adaptation scheme and integration into the transmission scheme.

#### **4.4.2 TACS performance evaluation**

In this section we are setting the main parameters to evaluate the performance of the TACS scheme, the IEEE 802.16 standard is here used to carry out the experiment. Some parameters are depicted in Table 1, where perfect synchronization is assumed and inter-cell interference is not considered. The used modulation is a Z-QAM (Z={2,4,16,64}) with Gray mapping. According to the CSI measured, the BS determines: *i*) the antenna subset, *ii*) the LDC subset and in case of throughput maximization, *iii*) the MCS that maximizes the rate for a maximum Block Error Rate - BLER (second optimization criterion). The codebook is composed mainly by the Single Input Multiple Output (Maximum Ratio Combining is used at the receiver) receiver, the Alamouti 's Spatial Diversity (SD) coding scheme (referred as G2 in hereafter plotted the figures) [15], the pure spatial multiplexing (SM) and the Golden code [17]. The performance of the system is evaluated over 100.000 channel realizations,

Space-Time Adaptation and MIMO Standardization Status 117

*gd=*(*N* – *M* + 1)=1. At higher data rates (*R*>8), all the codes perform similarly in the analysed

In Fig. 7 and Fig. 8, the bit error rate performance using TACS is shown having a fixed rate *R*=4. Fig. 7 shows the improvement due to the increase in *Ma* and also the performance achieved when combined with code selection. It can be observed how the TAS increases the diversity order, leading to a large performance increase for the SM and Golden subsets. It is very important to notice that despite the diversity increase for all the LDC subsets, SD and SIMO schemes still perform better when each code is evaluated independently. However, in Fig. 8, we can observe that when the code selection is switched on, SIMO and Golden subsets are selected most times, while the usage of SIMO increases with the SNR and the usage of SM and the Golden code increases with *Ma*. Furthermore, the achieved improvement by the TACS is clearly appreciated in Fig. 7, where an SNR improvement of approximately 1dB is obtained for *Ma*={3,4}. It is also surprising that the SM code is rarely selected knowing that the Golden code should always outperform SM since it obtains a higher diversity. However, as it is observed in Fig. 8, for less than 5% of the channel realizations the SM may outperform slightly the Golden code. Whether the singular value decomposition of the effective channel H is analysed when SM is selected, it has been observed that when all singular values are very close, both the SM and the Golden code lead

In Fig. 9 and Fig. 10, the performance using the TACS is again analysed for *R*=8. In Fig. 9 the different diversity orders of SD, SM, and the Golden Code are illustrated. We can appreciate here that the SM and the Golden code show the best performance when *Ma*={3,4}, and also for *Ma*=2 when SNR≤18dB. Furthermore the increase in the diversity order due to TACS can be observed in both Fig. 7 and Fig. 9. The maximum diversity order (*gd* = *MaN*) is achieved since at least one LDC (SIMO and G2) from those in the codebook are able to achieve the

Moreover, the BER using the TACS is equivalent to that obtained from the SISO scheme (referred as SISO*eq* in the plots) over a Rayleigh fading channel with the same rate *R,* a diversity order *gd*=*MaN* and a coding gain equal to . The performance of this *equivalent* SISO scheme, in terms of the bit error rate probability *Pb*, can be obtained directly by close

> *b b i P P i Z*

*P i kiZ k d Z Z*

 

1

2 1 <sup>3</sup> ,, 1 2 1

2 1

0 2

*Z*

<sup>1</sup> 2 1 ,, 1 2 2 (30)

log (28)

*b*

2 1 sin (29)

*Z*

/2 2

<sup>2</sup> log

*<sup>i</sup> <sup>d</sup> Z g*

*<sup>i</sup> k i <sup>i</sup> <sup>Z</sup> <sup>k</sup> kiZ*

<sup>1</sup> 2 1

expressions that are found in [41][42] and applying the Craig's formula in [43],

SNR range despite of the different diversity order between them.

**4.4.4 TACS performance under bit error rate minimization criterion** 

to very similar performances, therefore no matter which one is selected.

maximum diversity order.

*b*

*k*

0

12 1


Table 1. TACS evaluation framework system parameters

where for each realization a tile or a subchannel (specified in each analysis) is transmitted. In case of Partial Usage Subcarrier permutation (PUSC), the tile is formed by 4 subcarriers and 3 symbols, where 4 tones are dedicated to pilots as defined in IEEE 802.16e [17]. For the Band Adaptive Modulation and Coding (AMC) permutation scheme, each bin (equivalent to the tile concept) is comprised by 9 subcarriers where 1 tone is used as pilot. Perfect channel estimation is assumed at the receiver. Every *log2*(Z) bits are mapped to one symbol. The channel models used are uncorrelated Rayleigh (**H**~CN(0,1)) and the ITU Pedestrian A [38]. In both cases the channel is considered constant within a tile (*block fading channel model*). In case of uncorrelated Rayleigh the channel between tiles is uncorrelated, whereas in the ITU PedA case the channel is correlated both in frequency and time.

#### **4.4.3 MIMO reference and simulation results**

In Fig. 6, the reference performance for a fixed rated is depicted for *N*=2 when no transmit antenna selection neither code selection are used. For uncorrelated Rayleigh channel, we can observe that for low data rate, i.e. *R*={2,4}, the Alamouti code outperforms the rest of the schemes. This is strictly related to the diversity order that G2 achieves equal to *gd=N×M*=4, whereas the SM and the Golden code with a linear receiver get a diversity order of

<sup>1</sup> Forward error correction is consider only for the throughput maximization case, where the LUT used to predict the BLER as a function of the ESINR, are obtained using the Duo-Binary Turbo code defined for IEEE 802.16e.

Subcarrier Permutation Distributed (PUSC) and Contiguous (Band AMC)

Channel coding1 Turbo coding with rates: 1/3, 1/2, 2/3, ¾

Channel model Rayleigh and ITU Pedestrian A

Rate (spectral efficiency) {2,4,8} bits per channel use (bpcu)

where for each realization a tile or a subchannel (specified in each analysis) is transmitted. In case of Partial Usage Subcarrier permutation (PUSC), the tile is formed by 4 subcarriers and 3 symbols, where 4 tones are dedicated to pilots as defined in IEEE 802.16e [17]. For the Band Adaptive Modulation and Coding (AMC) permutation scheme, each bin (equivalent to the tile concept) is comprised by 9 subcarriers where 1 tone is used as pilot. Perfect channel estimation is assumed at the receiver. Every *log2*(Z) bits are mapped to one symbol. The channel models used are uncorrelated Rayleigh (**H**~CN(0,1)) and the ITU Pedestrian A [38]. In both cases the channel is considered constant within a tile (*block fading channel model*). In case of uncorrelated Rayleigh the channel between tiles is uncorrelated, whereas in the ITU

In Fig. 6, the reference performance for a fixed rated is depicted for *N*=2 when no transmit antenna selection neither code selection are used. For uncorrelated Rayleigh channel, we can observe that for low data rate, i.e. *R*={2,4}, the Alamouti code outperforms the rest of the schemes. This is strictly related to the diversity order that G2 achieves equal to *gd=N×M*=4, whereas the SM and the Golden code with a linear receiver get a diversity order of

1 Forward error correction is consider only for the throughput maximization case, where the LUT used to predict the BLER as a function of the ESINR, are obtained using the Duo-Binary Turbo code defined

Table 1. TACS evaluation framework system parameters

PedA case the channel is correlated both in frequency and time.

**4.4.3 MIMO reference and simulation results** 

for IEEE 802.16e.

Channel estimation (CQI) Ideal without any delay

OFDMA Air Interface and System Level configuration

FFT length, CP 2048, 12.5%

Modulation {4,16,64}-QAM
