**Adaptive Technique for Underwater Acoustic Communication**

Shen Xiaohong, Wang Haiyan, Zhang Yuzhi and Zhao Ruiqin *College of Marine Engineering, Northwestern Polytechnical University, Xi'an, China* 

## **1. Introduction**

18 Will-be-set-by-IN-TECH

58 Underwater Acoustics

Prada, C. & Fink, M. (1998). Separation of interfering acoustic scattered signals using the

Zhen, Y. (2001). Recent developments in underwater acoustics: Acoustic scattering from single and multiple bodies. *Proc. Natl.Sci. Counc. ROC(A)*, Vol.25, No.3, (2001)(137-150). Lim,R.; Lopes, J. L.; Hackman, R. H. & Todoroff, D. G.(1993). Scattering by objects buried in

Tesei, A.; Maguer, A. & Fox, W. L. J. (2002). Measurements and modeling of acoustic scattering

Junger, M. C.(1952). Sound scattering by thin elastic shells. *J. Acoust. Soc. Am.*, Vol. 24, No. 4,

Fawcett,J.A.; Fox, W.L. & Maguer, A. (1998). Modeling by scattering by objects on the seabed. *J. Acoust. Soc. Am.*, Vol. 104, No. 6, (dec. 1998) (3296-3304), ISSN 0001-4966. Wang, H. & Kaveh, M. (1985). Coherent signal-subspace processing for the detection and

Bourennane, S. & Bendjama, A. (2002). Locating wide band acoustic sources using higher-order statistics. *Applied Acoustics*, Vol. 63, No. 3, (march 2002) (235-251). Pillai, S.& Kwon, B. (1989). Forward/Backward spatial smoothing techniques for coherent

Frikel, M. & Bourennane, S. (1996). Fast algorithm for the wideband array processing using

Maheswara Reddy, K.& Reddy, V.U. (1996). Further results in spatial smoothing. *Signal*

Hung, H. & Kaveh, M. (1988). Focusing matrices for coherent signal-subspace processing.

*J. Acoust. Soc. Am.*, Vol. 104, No. 2, (august 1998).

(april 1993).

(july 1952).

5, (november 2002).

(1253-1263), ISSN 1053-587X .

Trieste Italy, Sept. 1996.

No. 1, (jan. 1989)(8-15), ISSN 0096-3518 .

*processing*, Vol. 48, (1996) (217-224).

(1272-1281), ISSN 0096-3518.

invariants of the time reversal operator. Application to Lamb waves characterization.

underwater sediments: Theory and experiment. *J. Acoust. Soc. Am.*, Vol. 93, No. 4,

from partially and completely buried spherical shells. *J. Acoust. Soc. Am.*, Vol. 112, No.

estimation of angles of arrival of multiple wide band sources. *IEEE Trans. on Acoustics, Speech and Signal Processing*, Vol. 33, No. 4, (aug. 1985) (823-831), ISSN 0096-3518. Yuen, N. & Friedlander, B. (1997). DOA estimation in multipath : an approach using

fourth order cumulants. *IEEE. Trans. on Signal Processing*, Vol. 45, No. 5, (may 1997)

signal identification. *IEEE Trans. on Acoustics, Speech and Signal Processing*, Vol. 37,

two-sided correlation transformation. *Proceedings of EUSIPCO'96*, Vol. 2, pp. 959-962,

*IEEE Trans. on Acoustics, Speech and Signal Processing*, Vol. 36, No. 8, (Aug. 1988)

Compared with the electromagnetic wave channel, the UWA (Underwater Acoustic) channel is characterized by large transmission delay, transmission loss increased with distance and frequency, serious multi-path effect, and remarkable Doppler Effect. These characteristics greatly influence the performance of UWA communication, and restrict channel capacity. Performance of UWA communication is far away from telecommunication, even if employ the similar techniques. Especially, available bandwidth is limited by spread distance because transmission loss increases with longer range distance. For example, the bandwidth for 5 km is 10 kHz whereas the bandwidth for 80 km is only 500Hz. For constant data rate, the bandwidth is confined by maximum working range and worst channel environment, so the utilization ratio of channel is very low. Besides the reasonable modulation model and advanced signal processing method which improves the detection performance of the receiver, the most fundamental way for reliable communication is adaptive modulation which greatly improves utilization ratio of bandwidth.

Modulation and detection techniques used for UWA communication include phase coherent (PSK and QAM) and noncoherent (FSK) techniques. The choice of modulation mode is based on the UWA channel parameters, such as multipath and the Doppler spread, as well as the SNR. The spread factor of the UWA channel determines whether phase coherent communications are possible. If so, an equalizer is employed to combat any intersymbol interference. If the channel varies too rapidly, noncoherent signaling is chosen. The choice of modulation is determined by the operator.

The objective of this chapter is based on the analysis of characteristics of UWA channel, developing algorithms that can self-adapting select the best technique for time varying UWA channel. Then the dynamic modulation and bandwidth optimization for high-rate UWA communication are deduced. At last the simulation results are shown.

## **2. Fading characteristics of UWA channel**

## **2.1 Characteristic of UWA channel**

Due to the absorption of medium itself, the wavefront expansion in sound propagation, the bending of acoustic ray, the scattering caused by various kinds of nonuniformity in ocean

Adaptive Technique for Underwater Acoustic Communication 61

( ) *<sup>r</sup> t <sup>P</sup> dB P*

Fig. 1. The relations between range and fading

( ) *<sup>r</sup> t*

*P dB P*

Fig. 2. The relations between time and fading

noise ratio) of receiver will be

**2.2 The relationship between bandwidth and frequency/ range in UWA channel** 

If only considering the large-scale fading, according to sonar function, the SNR(signal-to-

where *SL* is sound source level (dB), *NL* is noise spectrum level, B is the bandwidth. Assuming *NL*=45dB, the relationship between SNR of receiver and frequency (1~30kHz),

SNR *SL TL NL* 10logB (4)

and so on, the acoustic intensity will weaken in its propagation direction. The fading will change with distance, frequency and sensors location, which can be divided into large-scale fading and small-scale fading.

Large-scale fading: It is caused by sound propagation and sound ray convergence. The fading caused by sound propagation is composed by transmission loss and absorption loss, which is a function of range and frequency. It can be expressed as

$$TL = n \cdot 10 \log r + ar \tag{1}$$

where n=1,2,or 1.5, *r* is the range of communication, α is absorption coefficeent, which is a function of signal frequency *f*, and its value can be given by Thorp experience formula:

$$a(f) = \frac{0.11f^2}{1+f^2} + \frac{44f^2}{4100+f^2} + 3.0 \times 10^{-4} f^2 + 3.3 \times 10^{-3} \,\text{dB/km} \tag{2}$$

In the process of signal propagation, sound energy is strong in one area while weak in another area which is known as convergence fading.

Small-scale fading: small scale fading is composed by multipath and Doppler spread. The impulse response function is

$$h(\tau, t) = \sum\_{i=1}^{p} a\_i(t) \exp\left[j\phi\_i(t)\right] \delta\left[\tau - \tau\_i(t)\right] \tag{3}$$

where *P* is the number of multiple paths, *i* and *<sup>i</sup>* are the amplitude and phase of i-th path respectively. *<sup>i</sup>* is the time delay of the i-th path, and is uniformly distributed between min and max , where max is maximum delay spread . This kind of spread in time domain is corresponding to frequency selective decline in frequency domain, and the relation between coherence bandwidth *Bcoh* and maximum relative delay is max 1 *Bcok* . Multipath fading which caused by interference of multipath signals is related to the frequency of the signal, physical characteristics of the ocean, spatial location of the transmitter and the receiver.

Doppler Effect is caused by the relative motion between the transmitter and the receiver or the medium flow in UWA channel. Because multiple paths spread through different tracks, the received Doppler frequencies are also different from each other. Therefore the received signal is frequency spreading and the spread is measured by parameter *<sup>d</sup>* max *f fv c* , where max is the maximum relative radial velocity of multiple paths, c is the propagation velocity of sound waves, and *f* is the frequency of the signal transmission. Doppler spread in frequency domain is corresponding to time selective decline in time domain. Time selective decline is measured by relative time 1 *T f cok d* .The larger the relative time is, the slower channel varies. Conversely, the smaller the relative time is, the quicker channel varies.

By the above analysis, when the sea area, positions of the transmitter and of the receiver are established, large-scale fading is only the function of distance and frequency, and small scale fading is a random function of distance and time. In the adaptive UWA communication, big scale fading is slow fading, and small scale fading is fast fading. The relations between range and fading are shown in Fig.1. With defined distance, the relations between time and fading are shown in Fig.2.

60 Underwater Acoustics

and so on, the acoustic intensity will weaken in its propagation direction. The fading will change with distance, frequency and sensors location, which can be divided into large-scale

Large-scale fading: It is caused by sound propagation and sound ray convergence. The fading caused by sound propagation is composed by transmission loss and absorption loss,

*TL n r r* 10log

where n=1,2,or 1.5, *r* is the range of communication, α is absorption coefficeent, which is a function of signal frequency *f*, and its value can be given by Thorp experience formula:

In the process of signal propagation, sound energy is strong in one area while weak in

Small-scale fading: small scale fading is composed by multipath and Doppler spread. The

( , ) exp ( ) ( )

corresponding to frequency selective decline in frequency domain, and the relation between

which caused by interference of multipath signals is related to the frequency of the signal, physical characteristics of the ocean, spatial location of the transmitter and the receiver.

Doppler Effect is caused by the relative motion between the transmitter and the receiver or the medium flow in UWA channel. Because multiple paths spread through different tracks, the received Doppler frequencies are also different from each other. Therefore the received signal is frequency spreading and the spread is measured by parameter *<sup>d</sup>* max *f fv c* , where

max is the maximum relative radial velocity of multiple paths, c is the propagation velocity of sound waves, and *f* is the frequency of the signal transmission. Doppler spread in frequency domain is corresponding to time selective decline in time domain. Time selective decline is measured by relative time 1 *T f cok d* .The larger the relative time is, the slower channel varies. Conversely, the smaller the relative time is, the quicker channel varies.

By the above analysis, when the sea area, positions of the transmitter and of the receiver are established, large-scale fading is only the function of distance and frequency, and small scale fading is a random function of distance and time. In the adaptive UWA communication, big scale fading is slow fading, and small scale fading is fast fading. The relations between range and fading are shown in Fig.1. With defined distance, the relations between time and

*i* and 

*j t t* 

is the time delay of the i-th path, and is uniformly distributed between

is maximum delay spread . This kind of spread in time domain is

*iii*

 

0.11 44 ( ) 3.0 10 3.3 10

4 2 3

(3)

*<sup>i</sup>* are the amplitude and phase of i-th path

(1)

dB/km (2)

. Multipath fading

which is a function of range and frequency. It can be expressed as

2 2

1 4100

another area which is known as convergence fading.

2 2

*f f f f f f*

1

*i h t at* 

coherence bandwidth *Bcoh* and maximum relative delay is max 1 *Bcok*

*p*

fading and small-scale fading.

where *P* is the number of multiple paths,

, where max

impulse response function is

fading are shown in Fig.2.

respectively. *<sup>i</sup>*

 and max 

min 

Fig. 1. The relations between range and fading

Fig. 2. The relations between time and fading

#### **2.2 The relationship between bandwidth and frequency/ range in UWA channel**

If only considering the large-scale fading, according to sonar function, the SNR(signal-tonoise ratio) of receiver will be

$$\text{SNR} = \text{SL} - \text{TL} - \text{NL} - \text{10} \log \text{B} \tag{4}$$

where *SL* is sound source level (dB), *NL* is noise spectrum level, B is the bandwidth. Assuming *NL*=45dB, the relationship between SNR of receiver and frequency (1~30kHz),

Adaptive Technique for Underwater Acoustic Communication 63

, the interrupt threshold 0

,it can be proved when the optimal power allocation is

2

*H P*

*N B* 

 

  *j j*

(8)

*P P* 

(9)

, the transmitter won't transmit signal. This method is

1 1 0

,

 

. The capacity *C* is achieved by

 

, channel capacity achieves maximum as

<sup>0</sup> 2 0 *C B pd* log ( )

This capacity can be obtained by the time-varying transmission rates. Instantaneous SNR

**3.3 Channel capacity of time invariant frequency selective fading channel** 

*j j j*

*C B*

*P PP*

called frequency domain Power Water-filling Allocation method.

Interrupt threshold can be calculated by 11 1 <sup>0</sup> *<sup>j</sup>*

deteriorated, then the transmitter will reduce sending power and transmission rate. If the

Supposing the channel gain *H(f)* is block fading, the whole bandwidth can be divided into many sub-channels whose bandwidth are B, and in each sub-channel *H <sup>i</sup> f H* is constant. The SNR of i-th channel is <sup>2</sup> *H P NB i j* <sup>0</sup> , where *Pj* is the allocated power when *<sup>j</sup> <sup>j</sup> P P* . *P* is the upper limit of the total power. The channel capacity is the sum of the whole sub-

> 2 : 0 log 1

It can be proved when the optimal power allocation is 0 0

<sup>0</sup> 2 0 *C B pd* log ( )

*j* 

 

allocating different power and data rate for multiple sub-channels. When SNR of channel is admirable, distribute more power and use high transmission rate. If channel gets deteriorated, the transmitter will reduce sending power and transmission rate. If the SNR

In order to simplify the analysis, this chapter assumes that the response function of the UWA channel is *H (f, t),* and the bandwidth *B* can be divided into several sub-bandwidths

 

 

method is called time domain Power Water-filling Allocation method.

(7)

. This formula shows: If channel condition gets

, the transmitter won't transmit signal. This

can be calculated by the

0

 

If the sending power varies with

0 0

<sup>0</sup> <sup>0</sup> 1 1 () 1 *p d*

 

0

 

 

corresponds to the rate *B*log2 0

channel capacity achieves maximum as

falls below the interrupt threshold <sup>0</sup>

**3.4 Capacity of UWA channel** 

SNR falls below the interrupt threshold <sup>0</sup>

formula

1 1 0

*P P* 

channels

SNR of receiver and range(1~100km) are shown in Fig.3. From the figure we can see, if the transmitting power and the SNR of receiver are defined, the system bandwidth is the function of distance and frequency.

Fig. 3. The relationship between SNR and frequency, SNR and range

## **3. Channel capacity of UWA channel**

## **3.1 Channel capacity of AWGN (Additive White Gaussian Noise) time invariant channel**

Assuming *n(t)* is AWGN, 0 *N* 2 is PSD (Power Spectral Density) of AWGN, *P* is the average transmitted power, B is the receiving bandwidth, *C* is the channel capacity per second(bit/s). When the channel gain is constant 1, the SNR of the receiver is constant *t P NB* <sup>0</sup> , and then the channel capacity limited by average power can be expressed as

$$C = B \log\_2(1 + \gamma) \tag{5}$$

This formula shows the channel capacity is proportional to the bandwidth, and increases with the improvement of received SNR.

### **3.2 Channel capacity of time variant flat fading channel**

Supposing the gain of stationary and ergodic channel is *g t* , *g(t)* obeys distribution p(g), and is unrelated with channel input. So the instantaneous received SNR is <sup>0</sup> *t gtP NB* ( ) , the distribution of *t* is determined by *gt* , and the channel capacity is

$$C = \int\_0^\wp B \log\_2 \left( 1 + \gamma \right) p(\gamma) d\gamma \tag{6}$$

62 Underwater Acoustics

SNR of receiver and range(1~100km) are shown in Fig.3. From the figure we can see, if the transmitting power and the SNR of receiver are defined, the system bandwidth is the

Fig. 3. The relationship between SNR and frequency, SNR and range

**3.1 Channel capacity of AWGN (Additive White Gaussian Noise) time invariant** 

Assuming *n(t)* is AWGN, 0 *N* 2 is PSD (Power Spectral Density) of AWGN, *P* is the average transmitted power, B is the receiving bandwidth, *C* is the channel capacity per second(bit/s). When the channel gain is constant 1, the SNR of the receiver is

*C B* log 1 <sup>2</sup>

This formula shows the channel capacity is proportional to the bandwidth, and increases

Supposing the gain of stationary and ergodic channel is *g t* , *g(t)* obeys distribution p(g), and is unrelated with channel input. So the instantaneous received SNR

<sup>2</sup> <sup>0</sup>

*<sup>d</sup>*

 *p* 

*C B*log 1

*t P NB* <sup>0</sup> , and then the channel capacity limited by average power can be

(5)

*t* is determined by *gt* , and the channel

(6)

**3. Channel capacity of UWA channel** 

with the improvement of received SNR.

*t gtP NB* ( ) , the distribution of

**3.2 Channel capacity of time variant flat fading channel** 

**channel** 

constant

expressed as

is <sup>0</sup>

capacity is

function of distance and frequency.

If the sending power varies with , the interrupt threshold 0 can be calculated by the formula <sup>0</sup> <sup>0</sup> 1 1 () 1 *p d* ,it can be proved when the optimal power allocation is 0 0 0 1 1 0 *P P* , channel capacity achieves maximum as

$$C = \int\_{\gamma\_0}^{\infty} B \log\_2(\gamma/\gamma\_0) p(\gamma) d\gamma \tag{7}$$

This capacity can be obtained by the time-varying transmission rates. Instantaneous SNR corresponds to the rate *B*log2 0 . This formula shows: If channel condition gets deteriorated, then the transmitter will reduce sending power and transmission rate. If the SNR falls below the interrupt threshold <sup>0</sup> , the transmitter won't transmit signal. This method is called time domain Power Water-filling Allocation method.

#### **3.3 Channel capacity of time invariant frequency selective fading channel**

Supposing the channel gain *H(f)* is block fading, the whole bandwidth can be divided into many sub-channels whose bandwidth are B, and in each sub-channel *H <sup>i</sup> f H* is constant. The SNR of i-th channel is <sup>2</sup> *H P NB i j* <sup>0</sup> , where *Pj* is the allocated power when *<sup>j</sup> <sup>j</sup> P P* . *P* is the upper limit of the total power. The channel capacity is the sum of the whole subchannels

$$\mathbf{C} = \sum\_{p\_j:\sum\_j p\_j \le p} B \log\_2 \left( \mathbf{1} + \frac{\left| H\_j \right|^2 P\_j}{N\_0 B} \mathbf{\mathcal{V}} \right) \tag{8}$$

It can be proved when the optimal power allocation is 0 0 0 1 1 0 *P P* , channel capacity achieves maximum as

> <sup>0</sup> 2 0 *C B pd* log ( ) (9)

Interrupt threshold can be calculated by 11 1 <sup>0</sup> *<sup>j</sup> j* . The capacity *C* is achieved by allocating different power and data rate for multiple sub-channels. When SNR of channel is admirable, distribute more power and use high transmission rate. If channel gets deteriorated, the transmitter will reduce sending power and transmission rate. If the SNR falls below the interrupt threshold <sup>0</sup> , the transmitter won't transmit signal. This method is called frequency domain Power Water-filling Allocation method.

#### **3.4 Capacity of UWA channel**

In order to simplify the analysis, this chapter assumes that the response function of the UWA channel is *H (f, t),* and the bandwidth *B* can be divided into several sub-bandwidths

Adaptive Technique for Underwater Acoustic Communication 65

0

PSK 8.39 8.39 11.71 16.14 21.01

The UWA communication is power limited communication because of cavitation phenomenon and field effect. The large-scale fading is slow fading, and the small-scale

From analysis of section 3.4, adaptive modulation should be used to optimize the bandwidth ratio. By the use of multi-carrier transmission, the power of each sub-channel is allocated by SNR of receiver to achieve dynamic bandwidth. When SNR of channel is high, the transmitter distributes more power, or otherwise distributes less power. If the SNR falls

the transmission power and rate vary with the channel condition by Water-filling Allocation method. When SNR of channel is admirable, the transmitter distributes more power and uses higher transmission rate, or otherwise reduces sending power and transmission rate. If

In adaptive modulation system, at first, the transmitter sends a test signal to link the receiver, and then the receiver estimates the instantaneous channel characteristics and feeds back the updates to the transmitter, at last the transmitter selects a suitable modulation

  <sup>1</sup> *P P b B <sup>k</sup>* <sup>2</sup>

<sup>1</sup> *P P b B <sup>k</sup>* <sup>1</sup>

<sup>1</sup> 2 2 1 *k b B <sup>k</sup> P P* 

M=2 M=4 M=8 M=16 M=32

, the transmiter will not transmit signal. In one sub-channel

, do not transmit signal. It is dynamic

PSK 10.51 10.51 13.95 18.42 23.34 QAM 10.51 13.25 14.39 16.59 FSK 13.51 10.75 9.23 8.21 7.44

QAM 8.39 11.28 12.19 14.41 FSK 11.39 8.77 7.35 6.43 5.76

*<sup>W</sup>*

<sup>1</sup> log 2 *Rb <sup>M</sup>*

> 2 2 2

*Rb M M W M M* 

<sup>2</sup> *Rb* log *M W M*

log 4 log 4

SER BER bandwidth ratio

0 <sup>2</sup> 2 sin *<sup>b</sup>*

> ( 1) *b*

 

 

0

<sup>1</sup> <sup>1</sup> ( 1) <sup>1</sup> *<sup>M</sup> nk <sup>b</sup> n N n*

*M P e n n*

Table 1. SER, BER and bandwidth ratio of M-ary modulation models

Table 2. SNR to transmit 1bit information of M-ary modulation (dB)

the SNR falls below the interrupt threshold <sup>0</sup>

1 ( 1)

*<sup>k</sup> P Q N M*

<sup>3</sup> <sup>4</sup>

*<sup>k</sup> P Q M N* 

1

1

mode

**5. Adaptive UWA communication** 

*B*

*B*

*B n*

Modulation

PSK

QAM

FSK

(non-coherent detection)

<sup>6</sup> 10 *Pb*

<sup>4</sup> <sup>10</sup> *Pb*

fading is fast fading.

below the interrupt threshold <sup>0</sup>

modulation in one sub-carrier.

**5.1 Adaptive transmission system** 

by coherent bandwidth *Bcoh* . Each sub-channel is independent time variant flat fading channel, then for the j-th sub-channel, the response function is , *H ft H t <sup>j</sup>* . According to the allocated average power of each sub-channel, the capacity of every flat fading subchannel can be deduced. For sub-channels are independent from each other, the total power in time and frequency domain is the sum of capacity of each narrowband flat fading subchannel.

$$\mathcal{C} = \max\_{\{\overline{P}\_{j}\}: \sum\_{j} \overline{P}\_{j} \le \overline{P}} \sum\_{j} \mathcal{C}\_{j} \left(\overline{P}\_{j}\right) \tag{10}$$

where *C P j j* is the capacity of the sub-channel whose mean power is *P* and bandwidth is *Bcoh* . Bandwidth can be given by Eq. 6 From the two dimensions Water-filling Allocation method of time domain and frequency domain, when 0 0 0 1 1 0 *j j j j P P* , the channel capacity achieves maximum as

$$C = \sum\_{j} \int\_{\gamma\_0}^{\nu} B\_c \log\_2 \left( \gamma\_j / \gamma\_0 \right) p \left( \gamma\_j \right) d\gamma\_j \tag{11}$$

Interrupt threshold 0 can be calculated by <sup>0</sup> <sup>0</sup> 11 1 *j jj j p d* , which is the

same for each sub-channel.

To achieve the capacity in UWA communication, multi-carrier transmission should be used, and the power of each sub-channel is allocated by SNR. In one sub-channel, the transmission power and rate vary with the channel condition by Water-filling Allocation method.

## **4. Relationship between BER and SNR in digital communication**

PSK(Phase Shift Keying), QAM(Quadrature Amplitude Modulation) and FSK (Frequency Shift Keying) are the common modulation models used in UWA communication. Supposing the transmission signal ( ) *Ts t* , then the received signal with AWGN is

$$\mathbf{s}\_r(t) = \mathbf{s}\_T(t) + \mathbf{n}(t) \quad \text{ (} 0 \le t \le T \text{)}\tag{12}$$

where *n(t)* is AWGN sample with power spectrum density 0 () N 2 *nn f* . *N0* is the average power spectrum density of AWGN. With the optimized receiver, symbol error rate, bit error rate and bandwidth ratio of several modulation models are shown in Table1. *<sup>b</sup>* is signal power per bit.

According to Table1, when the BER is <sup>6</sup> 10 *Pb* or <sup>4</sup> <sup>10</sup> *Pb* , the SNR that M-ary modulation needs to transmit 1bit information is shown in Table2. The interrupt threshold in UWA communication is defined by the SNR.

64 Underwater Acoustics

by coherent bandwidth *Bcoh* . Each sub-channel is independent time variant flat fading channel, then for the j-th sub-channel, the response function is , *H ft H t <sup>j</sup>* . According to the allocated average power of each sub-channel, the capacity of every flat fading subchannel can be deduced. For sub-channels are independent from each other, the total power in time and frequency domain is the sum of capacity of each narrowband flat fading sub-

> : max *j j j*

 <sup>0</sup> 2 0 *<sup>c</sup>* log *j jj*

(10)

1 1

*j j j*

0

*p d*

, which is the

*P P*

(11)

 

0 0

*j*

 

(12)

, the SNR that M-ary modulation

is

0

, the

 

*j j P PP <sup>j</sup> C C P* 

where *C P j j* is the capacity of the sub-channel whose mean power is *P* and bandwidth is *Bcoh* . Bandwidth can be given by Eq. 6 From the two dimensions Water-filling Allocation

*C B pd*

 

 can be calculated by <sup>0</sup> <sup>0</sup> 11 1 *j jj j*

To achieve the capacity in UWA communication, multi-carrier transmission should be used, and the power of each sub-channel is allocated by SNR. In one sub-channel, the transmission power and rate vary with the channel condition by Water-filling Allocation

PSK(Phase Shift Keying), QAM(Quadrature Amplitude Modulation) and FSK (Frequency Shift Keying) are the common modulation models used in UWA communication. Supposing

> ( ) ( ) ( ) (0 ) *r T s t s t nt t T*

where *n(t)* is AWGN sample with power spectrum density 0 () N 2 *nn f* . *N0* is the average power spectrum density of AWGN. With the optimized receiver, symbol error rate, bit error rate and bandwidth ratio of several modulation models are shown in Table1. *<sup>b</sup>*

needs to transmit 1bit information is shown in Table2. The interrupt threshold in UWA

or <sup>4</sup> <sup>10</sup> *Pb*

method of time domain and frequency domain, when

*j*

**4. Relationship between BER and SNR in digital communication** 

the transmission signal ( ) *Ts t* , then the received signal with AWGN is

channel capacity achieves maximum as

Interrupt threshold 0

signal power per bit.

According to Table1, when the BER is <sup>6</sup> 10 *Pb*

communication is defined by the SNR.

method.

same for each sub-channel.

channel.


Table 1. SER, BER and bandwidth ratio of M-ary modulation models


Table 2. SNR to transmit 1bit information of M-ary modulation (dB)

## **5. Adaptive UWA communication**

The UWA communication is power limited communication because of cavitation phenomenon and field effect. The large-scale fading is slow fading, and the small-scale fading is fast fading.

From analysis of section 3.4, adaptive modulation should be used to optimize the bandwidth ratio. By the use of multi-carrier transmission, the power of each sub-channel is allocated by SNR of receiver to achieve dynamic bandwidth. When SNR of channel is high, the transmitter distributes more power, or otherwise distributes less power. If the SNR falls below the interrupt threshold <sup>0</sup> , the transmiter will not transmit signal. In one sub-channel the transmission power and rate vary with the channel condition by Water-filling Allocation method. When SNR of channel is admirable, the transmitter distributes more power and uses higher transmission rate, or otherwise reduces sending power and transmission rate. If the SNR falls below the interrupt threshold <sup>0</sup> , do not transmit signal. It is dynamic modulation in one sub-carrier.

## **5.1 Adaptive transmission system**

In adaptive modulation system, at first, the transmitter sends a test signal to link the receiver, and then the receiver estimates the instantaneous channel characteristics and feeds back the updates to the transmitter, at last the transmitter selects a suitable modulation

Adaptive Technique for Underwater Acoustic Communication 67

communication range. Supposing the distance is *r*, UWA velocity is 1500m/s, time delay 2 *di rc* , when 0.001 2 *<sup>d</sup> d cf* , system chooses the model to adapt small-scale fading,

Adaptive dynamic modulation and bandwidth optimization are powerful techniques to improve the energy efficiency and bandwidth ratio over UWA fading channel. Adaptive frequency and bandwidth are natural choices over UWA fading channel. Adaptive modulation and power management over UWA fading channel is investigated in this

For any UWA communication system, the most important parameters are carrier frequency and bandwidth, which are mainly dependent on range and ambient noise. Receiver expects a high SNR, so the principle to choose a proper carrier frequency and bandwidth is to maximize SNR. Based on sonar equation, the optimal frequency can be

> 70.7 ( ) *d FM <sup>f</sup> r df*

in which*, FM* is defined as *FM=SL-(NL-DI+DT)*, *SL* is the Source Level of transmitter, *NL* is water ambient Noise Level. The 3dB bandwidth of the optimistic frequency is the system

For double spread UWA channel, the spread factor is the product of time spread and frequency spread. If the spread factor accesses or overs 1, the channel will be overspread. Non-coherent modulation FSK has been considered as the only alternative for overspread

According the estimated distance *d* and Doppler frequency *fd*, if 0.001 2 *<sup>d</sup> d cf* , then the system will adapt to small-scale fading that is fast fading, otherwise will adapt to large-

SNR is an important parameter to optimize modulation and bandwidth in two dimensions channel. There are two estimation methods, one is estimating by sending the test training sequence, and the other is estimating by direct signals received. The former one is inefficient for lower bandwidth availability, and the later one is the main method of SNR estimation.

Supposing the transmission signal is ( ) *Ts t* , then the received signal through UWA channel

1/2

(14)

otherwise to adapt large-scale fading.

section.

determined by

bandwidth.

channel.

scale fading.

with AWGN is

**5.3.3 SNR estimation** 

**5.3 Dynamic modulation and bandwidth optimization** 

**5.3.1 Define carrier frequency and bandwidth according to the range** 

0

**5.3.2 Define the modulation model and adaptive fading model** 

scheme and bandwidth to transmit information. The receiver estimates channel and feeds back the updates to transmitter at the end of a frame. The adaptive system process block diagram is shown in Fig.4. The estimated received SNR of j-th channel at i-th time is <sup>0</sup> ˆ ˆ *j jj j i P g i NB* . *g i* ˆ *<sup>j</sup>* is the estimation of power gain of j-th channel. The relevant parameters are regulated based on estimated value at the time of integral times of code period *Ts*.

Fig. 4. Block diagram of adaptive system

#### **5.2 The adaptability criterion of large-scale fading and small-scale fading**

Supposing the transmitter sends information to the receiver at time i, after time delay ie, the receiver receivs the estimated SNR ˆ*ji* of j-th sub-channel at time i. Finally, with the additional feed back time delay *<sup>f</sup> i* , the receiver gets the information after time delay *d e <sup>f</sup> i ii* . Assuming the channel estimation and feed back information are both errorless, then the instantaneous BER is *P P b b* , and supposing *g*ˆ*i i <sup>d</sup>* is the actual value, the total BER of QAM is

$$P\_b\left(\mathcal{Y}\left[\mathbf{i}\right], \mathcal{Y}\left[\mathbf{i} - \mathbf{i}\_d\right]\right) \le 0.2 \left[\mathsf{S}P\_b\right]^{c\left[i, i\_d\right]}\tag{13}$$

It can be improved when Doppler frequency shift *fd* meets 0.001 *d e <sup>f</sup> <sup>d</sup> i ii f* , the average BER approaches desired value *Pb* . That is to say, the rate of channel variation effects the updates rate and estimated error of *i* .

As mentioned above, there are two aspects affecting channel gain, large-scale fading and small-scale fading. When Doppler frequency shift is small, the large-scale fading is constant and the small-scale fading varies slowly. In another word, in short range, estimated error and feed back delay can not weaken the system. When Doppler frequency shift is large, small-scale fading varies severely, that is system can not estimate channel and feed back information effectively in long range. Different adaptive control methods should be adopted according to different range, and there is a definite relation between control methods and 66 Underwater Acoustics

scheme and bandwidth to transmit information. The receiver estimates channel and feeds back the updates to transmitter at the end of a frame. The adaptive system process block diagram is shown in Fig.4. The estimated received SNR of j-th channel at i-th time is

parameters are regulated based on estimated value at the time of integral times of code

*g*[ ]*i n i*[ ]

*<sup>R</sup>*[ ]*<sup>i</sup> <sup>P</sup>*[ ]*<sup>i</sup> r i*[ ] *<sup>x</sup>*[ ]*<sup>i</sup> y i*[ ]

ˆ[ ] *e f g ii i*

**5.2 The adaptability criterion of large-scale fading and small-scale fading** 

 

> *i* .

Supposing the transmitter sends information to the receiver at time i, after time delay ie, the

additional feed back time delay *<sup>f</sup> i* , the receiver gets the information after time delay *d e <sup>f</sup> i ii* . Assuming the channel estimation and feed back information are both

, , 0.2 5 *<sup>d</sup> i i P i ii P b db*

It can be improved when Doppler frequency shift *fd* meets 0.001 *d e <sup>f</sup> <sup>d</sup> i ii f* , the average BER approaches desired value *Pb* . That is to say, the rate of channel variation effects the

As mentioned above, there are two aspects affecting channel gain, large-scale fading and small-scale fading. When Doppler frequency shift is small, the large-scale fading is constant and the small-scale fading varies slowly. In another word, in short range, estimated error and feed back delay can not weaken the system. When Doppler frequency shift is large, small-scale fading varies severely, that is system can not estimate channel and feed back information effectively in long range. Different adaptive control methods should be adopted according to different range, and there is a definite relation between control methods and

ˆ*ji* of j-th sub-channel at time i. Finally, with the

(13)

, and supposing *g*ˆ*i i <sup>d</sup>* is the actual

Fig. 4. Block diagram of adaptive system

receiver receivs the estimated SNR

value, the total BER of QAM is

updates rate and estimated error of

errorless, then the instantaneous BER is *P P b b*

ˆ *<sup>j</sup>* is the estimation of power gain of j-th channel. The relevant

*r i* ˆ[ ]

*i* ˆ[ ]*<sup>e</sup> g i i*

*e*

<sup>0</sup> ˆ ˆ *j jj j*

*i P g i NB* . *g i*

period *Ts*.

communication range. Supposing the distance is *r*, UWA velocity is 1500m/s, time delay 2 *di rc* , when 0.001 2 *<sup>d</sup> d cf* , system chooses the model to adapt small-scale fading, otherwise to adapt large-scale fading.

## **5.3 Dynamic modulation and bandwidth optimization**

Adaptive dynamic modulation and bandwidth optimization are powerful techniques to improve the energy efficiency and bandwidth ratio over UWA fading channel. Adaptive frequency and bandwidth are natural choices over UWA fading channel. Adaptive modulation and power management over UWA fading channel is investigated in this section.

## **5.3.1 Define carrier frequency and bandwidth according to the range**

For any UWA communication system, the most important parameters are carrier frequency and bandwidth, which are mainly dependent on range and ambient noise. Receiver expects a high SNR, so the principle to choose a proper carrier frequency and bandwidth is to maximize SNR. Based on sonar equation, the optimal frequency can be determined by

$$f\_0 = \left(\frac{70.7}{r} \cdot \frac{d(FM)}{df}\right)^{1/2} \tag{14}$$

in which*, FM* is defined as *FM=SL-(NL-DI+DT)*, *SL* is the Source Level of transmitter, *NL* is water ambient Noise Level. The 3dB bandwidth of the optimistic frequency is the system bandwidth.

### **5.3.2 Define the modulation model and adaptive fading model**

For double spread UWA channel, the spread factor is the product of time spread and frequency spread. If the spread factor accesses or overs 1, the channel will be overspread. Non-coherent modulation FSK has been considered as the only alternative for overspread channel.

According the estimated distance *d* and Doppler frequency *fd*, if 0.001 2 *<sup>d</sup> d cf* , then the system will adapt to small-scale fading that is fast fading, otherwise will adapt to largescale fading.

#### **5.3.3 SNR estimation**

SNR is an important parameter to optimize modulation and bandwidth in two dimensions channel. There are two estimation methods, one is estimating by sending the test training sequence, and the other is estimating by direct signals received. The former one is inefficient for lower bandwidth availability, and the later one is the main method of SNR estimation.

Supposing the transmission signal is ( ) *Ts t* , then the received signal through UWA channel with AWGN is

Adaptive Technique for Underwater Acoustic Communication 69

OFDM transports a signal-input data stream on several carriers within the usable frequency band of the channel. This is accomplished by partitioning the entire channel into *N* parallels, ideally orthogonal, and spectrally flat subchannels, each of equal bandwidth, and with center frequency *fn*, *n=1…N*. Thus an OFDM symbol is consisted of several subcarriers which are modulated as PSK or QAM. Each subcarrier can be independently modulated in adaptive modulation schemes or all subcarriers may be modulated in same manner. The

j2π

*f t*

where *Ts* is the symbol period, *f0* is the carrier frequency, *N* is the number of subcarriers, *d(n)* is the PSK or QAM symbol, and *t0* is the symbol of starting time. The bandpass signal in

<sup>2</sup><sup>π</sup> ( ) [ ( )exp(j )]exp(j2<sup>π</sup> ) ( ) exp(j2<sup>π</sup> )

The baseband signal *x(t)* is sampled at a rate of *fs*, then *tk=k/fs*. The baseband signal *X(t)* can

*N*

*N*

Noncoherent detection MFSK transports a signal-input data stream on selected carriers among *N* parallel channels based on coding. A MFSK symbol can be considered as a special

*others*

It is clear from Eq. 22, 23 and 24 that dynamic adaptive modulation or demodulation combines MFSK and OFDM effectively and employs the FFT/IFFT algorithm to synthesize the modulation or demodulation without any other algorithms. It selects a suitable modulation or demodulation between OFDM and MFSK according to the estimated

*if sending subcarrier*

where *x(k)=x(tk)*. Eq. 22 shows that *x(t)* is the Inverse Fast Fourier Transform (IFFT) of *d(n)*. At receiver the FFT is applied to the discrete time OFDM signal *x(t)* to recovery the *d(n)* 

*x t d n nt f t X t f t <sup>T</sup>*

0 0 [, ]*<sup>s</sup> t tt T* (20)

0 0

, 0 ( 1) *k N* (22)

, 0 ( 1) *n N* (23)

(24)

(21)

**5.4 The adaptive multi-carrier modulation in UWA communication** 

1

0 ( ) ( )e , *<sup>o</sup> N*

*n xt dn* 

bandpass OFDM symbol can be expressed as follows:

1

*N*

0

*n s*

1

*N*

*n*

<sup>2</sup><sup>π</sup> ( ) ( )exp(j )

<sup>2</sup><sup>π</sup> ( ) ( )exp( j )

*d n X k nk*

*X k d n nk*

0

1

*N*

*k*

0

case of OFDM symbols. The condition of *d(n)* to be satisfied is that

*d n*

parameters of UWA communication channel.

1, ( ) 0,

Eq. 20 can also be expressed in the form of

be expressed in the form of

symbols written in

$$y(t) = \sum\_{l=1}^{p} \alpha\_l \exp\left[j(\phi\_l(t))s\_T(t-\tau\_l f\_o) + n(t)\right] \tag{15}$$

The self-correlation function of the received signal is

$$\begin{aligned} \phi\_{yy}(\tau) &= \mathbb{E}[y(t)y(t+\tau)] \\ &= \mathbb{E}[y\_I(t)y\_I(t+\tau)]\cos 2\pi f\_c \tau \\ &- \mathbb{E}[y\_Q(t)y\_I(t+\tau)]\sin 2\pi f\_c \tau + \sigma\_n^2 \delta(\tau) \\ &= \phi\_{ylyl}(\tau)\cos 2\pi f\_c \tau - \phi\_{ylyl}(\tau)\sin 2\pi f\_c \tau + \sigma\_n^2 \delta(\tau) \\ &= \sigma^2 f\_0(2\pi f\_m \tau)\cos(2\pi f\_c \tau) + \sigma\_n^2 \delta(\tau) \end{aligned} \tag{16}$$

In the formula, <sup>2</sup> is signal power, <sup>2</sup> *<sup>n</sup>* is noise power, 0*J x*( ) is the zero-order Bessel function of first kind, *<sup>m</sup>* / <sup>0</sup> *f ff* is normalization Doppler frequency. From the properties of channel impulse response, the above formula is discretized and normalized as

$$\varphi\_{yy}(kT\_s) = \frac{\sigma^2 f\_0(2\pi f\_m kT\_s) \cos(2\pi f\_c kT\_s)}{\sigma^2 + \sigma\_n^2} \quad (k = 1, 2, 3...) \tag{17}$$

According to the definition of SNR

$$\text{SNR} = \frac{\rho\_{yy}(kT\_s)}{I\_o(2\pi f\_m kT\_s)\cos(2\pi f\_c kT\_s) - \rho\_{yy}(kT\_s)} \qquad \text{(\$k=1,2,3...)}\tag{18}$$

If *fm* is unknown, the equation set can be deduced from Eq.18

$$\begin{cases} \mathcal{Y} = \frac{\rho\_{yy}(T\_s)}{J\_0(2\pi f\_m T\_s)\cos(2\pi f\_c T\_s) - \rho\_{yy}(T\_s)}\\ \mathcal{Y} = \frac{\rho\_{yy}(2T\_s)}{J\_0(4\pi f\_m T\_s)\cos(4\pi f\_c T\_s) - \rho\_{yy}(2T\_s)} \end{cases} \tag{19}$$

 can be calculated by the equation set. If the system adapts to slow fading channel, the estimated can be averaged.

#### **5.3.4 Allocate the instantaneous power of each sub-channel based on the estimated SNR**

First, the power of each sub-channel is allocated by 0 0 0 1 1 0 *j j j j P P* . Then,

instantaneous power of each sub-channel is allocated by 0 0 0 1 1 0 *P P* 

68 Underwater Acoustics

( ) exp[ ( ( )] ( ) ( )

[ ( ) ( )]cos2

*Ey ty t f Ey ty t f*

*Jf f*

of channel impulse response, the above formula is discretized and normalized as

 

*yy s o m c s yy s kT SNR k J f kT f kT kT <sup>s</sup>*

 

*I I c*

2 2

[ ( ) ( )]sin 2 ( )

 

*Q I c n*

(2 )cos(2 ) ( )

 

*m cn*

function of first kind, *<sup>m</sup>* / <sup>0</sup> *f ff* is normalization Doppler frequency. From the properties

 

( ) ( 1,2,3...) (2 )cos(2 ) ( )

( ) (2 )cos(2 ) ( ) (2 ) (4 )cos(4 ) (2 )

*T J f T fT T*

*T J f T fT T*

*yy s m s c s yy s yy s m s c s yy s*

can be calculated by the equation set. If the system adapts to slow fading channel, the

**5.3.4 Allocate the instantaneous power of each sub-channel based on the estimated** 

instantaneous power of each sub-channel is allocated by 0 0

(18)

2 2 (2 )cos(2 ) ( ) 1,2,3... *ms cs n*

*J f kT f kT kT <sup>k</sup>*

*yIyI c yIyQ c n*

 

 

*l l T lo*

 *j ts t f n t*

2

 

( )cos2 ( )sin 2 ( )

*f f*

(15)

2

*<sup>n</sup>* is noise power, 0*J x*( ) is the zero-order Bessel

(17)

(16)

(19)

0 0

*j*

 

 

1 1

*j j j*

1 1 0

 0

*P P*

*P P*  0

 

0

. Then,

 

1

*l*

( ) [ ( ) ( )]

*Eytyt*

 

0

is signal power, <sup>2</sup>

2 0

 

*y t*

The self-correlation function of the received signal is

*yy*

 

*yy s*

First, the power of each sub-channel is allocated by

can be averaged.

If *fm* is unknown, the equation set can be deduced from Eq.18

0

0

According to the definition of SNR

In the formula, <sup>2</sup>

**SNR** 

estimated

*p*

#### **5.4 The adaptive multi-carrier modulation in UWA communication**

OFDM transports a signal-input data stream on several carriers within the usable frequency band of the channel. This is accomplished by partitioning the entire channel into *N* parallels, ideally orthogonal, and spectrally flat subchannels, each of equal bandwidth, and with center frequency *fn*, *n=1…N*. Thus an OFDM symbol is consisted of several subcarriers which are modulated as PSK or QAM. Each subcarrier can be independently modulated in adaptive modulation schemes or all subcarriers may be modulated in same manner. The bandpass OFDM symbol can be expressed as follows:

$$\mathbf{x}(t) = \sum\_{n=0}^{N-1} d(n) \mathbf{e}^{\mathbf{j}2\pi f\_s t}, \quad t \in [t\_0, t\_0 + T\_s] \tag{20}$$

where *Ts* is the symbol period, *f0* is the carrier frequency, *N* is the number of subcarriers, *d(n)* is the PSK or QAM symbol, and *t0* is the symbol of starting time. The bandpass signal in Eq. 20 can also be expressed in the form of

$$\mathbf{x}(t) = \left[\sum\_{n=0}^{N-1} d(n) \exp(j \frac{2\pi}{T\_s} nt)\right] \exp(j2\pi f\_0 t) = X(t) \times \exp(j2\pi f\_0 t) \tag{21}$$

The baseband signal *x(t)* is sampled at a rate of *fs*, then *tk=k/fs*. The baseband signal *X(t)* can be expressed in the form of

$$X(k) = \sum\_{n=0}^{N-1} d(n) \exp(\text{j}\frac{2\pi}{N}nk) \text{ , } 0 \le k \le (N-1) \tag{22}$$

where *x(k)=x(tk)*. Eq. 22 shows that *x(t)* is the Inverse Fast Fourier Transform (IFFT) of *d(n)*.

At receiver the FFT is applied to the discrete time OFDM signal *x(t)* to recovery the *d(n)*  symbols written in

$$d(n) = \sum\_{k=0}^{N-1} X(k) \exp(-j\frac{2\pi}{N}nk), \; 0 \le n \le (N-1) \tag{23}$$

Noncoherent detection MFSK transports a signal-input data stream on selected carriers among *N* parallel channels based on coding. A MFSK symbol can be considered as a special case of OFDM symbols. The condition of *d(n)* to be satisfied is that

$$d(n) = \begin{cases} 1, \text{ if } \text{sending subcarrier} \\ 0, \text{ others} \end{cases} \tag{24}$$

It is clear from Eq. 22, 23 and 24 that dynamic adaptive modulation or demodulation combines MFSK and OFDM effectively and employs the FFT/IFFT algorithm to synthesize the modulation or demodulation without any other algorithms. It selects a suitable modulation or demodulation between OFDM and MFSK according to the estimated parameters of UWA communication channel.

Adaptive Technique for Underwater Acoustic Communication 71

From the figure, we can see the SNR estimated algorithm performance well. The standard deviation of estimated SNR is less than 0.7dB. So the estimation is available in adaptive

The parameters of adaptive UWA communication system used for performance simulation are showed in Table 3. A linear frequency modulation signal is used for synchronization.

MFSK 2~2.2kHz 200Hz 0s 0.2s 1

OFDM 2~5 kHz 3 kHz 0.05s 300

To overcome the ISI, system uses a grouped FSK modulation technique. The band is divided into 2 groups, only one of which is transmitted at one time. System uses multiple FSK modulation technique in a group. The band is divided into 2, 4 or 8 subbands based on

Fig. 7 and Fig.8 show that the bandwidth raises at the price of increasing SNR at the same error bit. Compared with Fig.3, 200Hz bandwidth 8FSK can be used for UWA communication at 100km range, and 1200Hz can be used for 40 km. OFDM with 3 kHz bandwidth can be used for UWA communication at 15 km range, and 6 kHz can be used for

An UWA communication experiment which using adaptive system with 1/2 rate turbo error control coding was taken in lake. The results demonstrate the performance of adaptive communication system at the range from 5km to 25km. The modulation schemes, the

number of subcarriers, data rate and the bit error rate are given in Table 4.

interval

2~2.4kHz 400Hz 2

2~2.8kHz 800Hz 4

2~3.6kHz 1600Hz 8

2~6 kHz 4 kHz 400

2~7 kHz 5 kHz 500

2~8 kHz 6kHz 600

FFT /IFFT Number of sub carrier

The UWA channel is based on the Fig.5, and the relative velocity is 1.5m/s.

Modulation Bandpass Bandwidth Guard

UWA communication.

Table 3. Adaptive UWA parameters

the short range.

**6.3 Lake experiments** 

bandwidth, in each of which an 8FSK signal is transmitted.

**6.2 Simulations** 

## **6. Simulation and experiment**

## **6.1 SNR estimation**

In order to evaluate the performance of the dynamic adaptive system, an UWA communication experiment was conducted in a lake. The impulse response of the lake is shown in Fig.5.

According to the SNR estimated method, the 2FSK, 2PSK and 4QAM signals in different SNRs are respectively simulated in 100 times. Considering the SNR varying on the receiver and practical application, with the range of SNR [-10 dB, 10 dB ], the estimated standarddeviations are shown in Fig.6

Fig. 5. Impulse response of the lake

Fig. 6. Average SNR

From the figure, we can see the SNR estimated algorithm performance well. The standard deviation of estimated SNR is less than 0.7dB. So the estimation is available in adaptive UWA communication.

## **6.2 Simulations**

70 Underwater Acoustics

In order to evaluate the performance of the dynamic adaptive system, an UWA communication experiment was conducted in a lake. The impulse response of the lake is

According to the SNR estimated method, the 2FSK, 2PSK and 4QAM signals in different SNRs are respectively simulated in 100 times. Considering the SNR varying on the receiver and practical application, with the range of SNR [-10 dB, 10 dB ], the estimated standard-

**6. Simulation and experiment** 

deviations are shown in Fig.6

Fig. 5. Impulse response of the lake

Fig. 6. Average SNR

**6.1 SNR estimation** 

shown in Fig.5.

The parameters of adaptive UWA communication system used for performance simulation are showed in Table 3. A linear frequency modulation signal is used for synchronization. The UWA channel is based on the Fig.5, and the relative velocity is 1.5m/s.


Table 3. Adaptive UWA parameters

To overcome the ISI, system uses a grouped FSK modulation technique. The band is divided into 2 groups, only one of which is transmitted at one time. System uses multiple FSK modulation technique in a group. The band is divided into 2, 4 or 8 subbands based on bandwidth, in each of which an 8FSK signal is transmitted.

Fig. 7 and Fig.8 show that the bandwidth raises at the price of increasing SNR at the same error bit. Compared with Fig.3, 200Hz bandwidth 8FSK can be used for UWA communication at 100km range, and 1200Hz can be used for 40 km. OFDM with 3 kHz bandwidth can be used for UWA communication at 15 km range, and 6 kHz can be used for the short range.

## **6.3 Lake experiments**

An UWA communication experiment which using adaptive system with 1/2 rate turbo error control coding was taken in lake. The results demonstrate the performance of adaptive communication system at the range from 5km to 25km. The modulation schemes, the number of subcarriers, data rate and the bit error rate are given in Table 4.

Adaptive Technique for Underwater Acoustic Communication 73

MFSK was seen as intrinsically robust for the time and frequency spreading of long range UWA channel. OFDM has been used in UWA communication at short or medium range. Adaptive UWA communication system combines MFSK and OFDM effectively, which dynamic selects modulation schema and optimizes bandwidth based on UWA communication estimation. This method has obvious advantages: being realized by DFT based filter banks as OFDM, good performance and the high frequency band efficiency in time varying fading UWA channel. Based on the results of simulation and experiments in a lake, it is shown that the adaptive UWA communication system is more efficient for high rate UWA communication not only at short range, but also at medium and long range.

Andrea Goldsmith. Wireless communications. Camgridge university press. 2005.

Oceanic Engineering, 2000; 25(1): 52-61.

communications in Brest. OCEANS 2005:425~430.

IEEE Trans. Commu.. 26(9):1754~1766,2008.

Oceanic Engineering, 1996; 21(2):181-191.

Journal of Oceanic Engineering, 1996; 21(2): 125-136.

Benson A., Proakis J., Stojanovic M., Towards robust adaptive acoustic communications[C]. OCEANS 2000 MTS/IEEE Conference and Exhibition. 2:1243 ~ 1249 ,2000. Bayan S. Sharif, Oliver R. Hinton, & Alan E. Adams. A Computationally Efficient Doppler

G.Lapierre, N.Beuzelin.1995-2005:Ten years of active research on underwater acoustic

Kilfoyle Daniel B., Baggeroer Arthur B. The State of the Art in Underwater Acoustic

K. F. Scussel, J. A. Rice, and S. Merriam. A new MFSK acoustic modem for operation in adverse underwater channels. Oceans'97, Halifax, NS, Canada, 1997. Michele Zorzi. Energy-Efficient Routing Schemes for Underwater Acoustic Networks[J].

Stojanovic M. Recent Advances in High-speed Underwater Acoustic Communications. IEEE

Woodward B, Sari H. Digital UnderwaterAcoustic Voice Communications. IEEE Journal of

Telemetry. IEEE Journal of Oceanic Engineering, 2000; 25(1): 4-27

Compensation System for Underwater Acoustic Communications. IEEE Journal of

5 OFDM 1000 9090 5.33 <10-4 10 OFDM 500 3400 8.40 <10-4 15 OFDM 200 1360 8.32 <10-4 25 8FSK 16 640 2.55 <10-4

data rate bs-1

BER

/% BER after decoding

subcarriers

Range

**7. Conclusion** 

**8. References** 

/km modulation number of

Table 4. Results of lake experiment

Fig. 7. BER Performance of four kinds of bandwidth of MFSK

Fig. 8. BER Performance of four kinds of bandwidth of OFMD


Table 4. Results of lake experiment

## **7. Conclusion**

72 Underwater Acoustics

Fig. 7. BER Performance of four kinds of bandwidth of MFSK

Fig. 8. BER Performance of four kinds of bandwidth of OFMD

MFSK was seen as intrinsically robust for the time and frequency spreading of long range UWA channel. OFDM has been used in UWA communication at short or medium range. Adaptive UWA communication system combines MFSK and OFDM effectively, which dynamic selects modulation schema and optimizes bandwidth based on UWA communication estimation. This method has obvious advantages: being realized by DFT based filter banks as OFDM, good performance and the high frequency band efficiency in time varying fading UWA channel. Based on the results of simulation and experiments in a lake, it is shown that the adaptive UWA communication system is more efficient for high rate UWA communication not only at short range, but also at medium and long range.

## **8. References**

Andrea Goldsmith. Wireless communications. Camgridge university press. 2005.


**5** 

*China* 

Weijie Shen, Haixin Sun\*,

En Cheng, Wei Su and Yonghuai Zhang

**Narrowband Interference Suppression in** 

In 1960, Chang [1] postulated the principle of transmitting messages simultaneously through a linear band-limited channel without ICI and ISI. Shortly after, Saltzberg [2] analyzed the performance of such a system and concluded, "The efficient parallel system needs to concentrate more on reducing cross talk between the adjacent channels rather than perfecting the individual channel itself because imperfection due to cross talk tends to dominate." This was an important observation and was proved in later years in the case of

The major contribution to the OFDM technique came to fruition when Weinstein and Ebert [3] demonstrated the use of the Discrete Fourier Transform (DFT) to perform the base-band modulation and demodulation. The use of the DFT immensely increased the efficiency of the modulation and demodulation processing. The use of the guard space and the raisedcosine filtering solve the problems of ISI to a great extent. Although the system envisioned as such did not attain the perfect orthogonality between sub-carriers in a time-dispersive channel, nonetheless it was still a major contribution to the evolution of OFDM systems.

In quest of solving the problem of orthogonality over the dispersive channel, Peled and Ruiz [4] introduced the notion of cyclic prefix (CP). They suggested filling the guard space with the cyclic extension of the OFDM symbol, which acts as if it is performing the cyclic convolution by the channel as long as the channel impulse response is shorter than the length of the CP, thus preserving the orthogonality of sub-carriers. Although addition of the CP causes a reduction of the data rate, this deficiency was more than compensated by the

Orthogonal Frequency Division Multiplexing (OFDM) scheme enables channel equalization in the frequency domain, thus eliminating the need for potentially complex time-domain equalization of a single-carrier system [5]. For this reason, OFDM has found application in a number of systems. Underwater acoustic(UWAC) OFDM system has been widely studied to overcome the complexity of underwater acoustic channels [6] and get a high speed data

**1. Introduction** 

baseband digital signal processing.

ease of receiver implementation.

Corresponding Author

 \*

**Underwater Acoustic OFDM System** 

*The Key Laboratory of Underwater Acoustic Communication and Marine Information Technology, Xiamen University, Xiamen, Fujian,* 

