**3. The WSN hardware architecture**

The hardware of wireless sensor motes consists of sensors (analog and/or digital), a microcontroller, also referred to as a microprocessor or Central Processing Unit(CPU), memory, RF communication module (transceiver) and battery. The design of each component of a WSN mote should take into consideration the power metrics (power consumption and voltage requirements) of the component. Additionally, the integration/interface of all the components as a whole should be studied for power consumption (having analog sensors means that an ADC component in the CPU should be required to convert the sensor readings to a digital format etcÉ)

To reduce power consumption, several works suggest the introduction of sleep and wake up cycles for the motes. Other schemes suggest a better integration of the functionality of hardware components (using cross-layer principles). Another consideration in the design of the CPU is the clock component. Several applications of WSNs require some level of time synchronization. Clock choices and designs affect the amount of drift that a sensor mote's clock can experience requiring more or less time synchronization operations when the mote is deployed (Akyildiz et al., 2002).

#### **3.1 Sensors**

Of the five main units, the sensing unit is the most application specific. Meaning the type of sensor used will depend on the application. For instance, wireless sensors used for structural health monitoring may consist of materials apt for monitoring strain, acceleration (accelerometer) and linear and angular displacement. Other application specific sensors may measure, vehicular movement, soil consistency, blood alcohol levels, humidity, noise levels and so on. These sensors should then report some signal indicative of their acquisition. Temperature (thermo-coupler outputting voltage or thermistor outputting resistance), force or pressure (piezoelectric outputting voltage, strain gauge outputting resistance), position (linear variable differential transformers (LVDT) outputting alternating current) or light intensity (photodiode outputting current) all need to report information regarding their surroundings to a processing unit (Wilson, 2004).

Fig. 2. Examples of a 3D optical metamaterials fabricated in a layer-by-layer manner. (a) A near-infrared NIM (Negative Index Material) with three functional layers made by EBL (Electron Beam Lithography); (b) Four layers of SRRs (Split Ring Resonators) based on EBL with patterning-and-flattening approach; (c) A NIM wedge exhibiting negative refraction for

Power Considerations for Sensor Networks 133

reflect but to divert the energy through a desired path. It is no wonder as to the attention metamaterials have seen for energy harvesting. Research is currently being conducted to develop sensors (both photon and electron based) that extract atmospheric energy regardless of the incident angle such that no energy is reflected back out of the sensor rather, it's reflected down toward the detector. This will lead to the creation of ultra-efficient sensors for wireless networks, see (Narimanov & Kildishev, 2009; Shalaev et al., 2005) for more information on

The component responsible for doing the bulk of the switching and decision making for the WSN at the remote site is the processor or microcontroller. When selecting the processor for specific WSN applications, the engineer must make many considerations. These considerations include, but are not limited to, cost, power requirements, physical size, weight

Depending on the microcontroller, the power requirement could range from .25 mA to 2.5 mA per MHz for either 8 or 16 bit processors. Again, the application will determine if a processor consuming relatively high amounts of energy is acceptable or if .25 mA per MHz is needed. A common misconception is that by putting the processor in "sleep" mode, the sensor utilizes less power thus is more efficient. It has been shown that this is not always true as while in "sleep mode", sensors still maintain synchronization and memory functionalities necessary to

In fact a more prudent approach to saving energy would include completely shutting the processor off, entirely, and ensuring the sensor can rapidly recover from a "dead" state or at the very least rapidly jump from "sleep" mode to "awake" mode. As the processor needs to synchronize native clocks and stabilize, the transition time or delay can be as long as 10 ms which is a relative eternity. Another parallel approach involves varying the speed depending

In other words, only using the minimum power required for a task at a given time by dynamically ramping up or down the power accordingly versus drawing full power for all "awake" states. This approach may benefit from an algorithm in which the speed is a function of the power. If the required task and itÕs effort expended is known before the task is given, an absolute "finish time" can be maintained without necessarily completing the task as fast

visible light made by an advanced FIB technique Cai &shaleav (2010).

and speed, some of which are elaborated upon below.

perform expeditiously upon awakening (Hu & Cao, 2010).

on the time allotted for a specific task.

Metamaterials.

**3.2 Microcontroller**

The development of an efficient method for acquiring and converting conventional energy from the sensors such as solar and wind has seen an exponential growth over the last few years. The sensing development has been referred to as energy harvesting. One factor contributing to the enjoyment of such an increase has been the threat of rapid decreases projected in our global and national energy reserves based on utilization rates and trends. Such a premonition has spurred funding for research in various fields including materials or more specifically metamaterials.

Metamaterials have been defined by most associated scientist as materials made by man which exhibit non-natural properties and characteristics, particularly EM or electromagnetic properties not known to exist with any other materials found in nature. Regarding electromagnetism, metamaterials which exhibit propagating electromagnetic waves (both the permittivity and permeability are negative as seen in Figure 1 have seen much attention in recent years as well as when both permittivity and permeability are very close to 1.

Fig. 1. The parameter space for *�* and *μ*. The two axes correspond to the real parts of permittivity and permeability, respectively. The dashed green line represents non-magnetic materials with *μ* = 1 Cai &shaleav (2010).

The reaction or response of a material (as in a sensor for WSNs) to external fields is largely determined by only the two material parameters *�* and *μ*, permittivity and permeability respectively. As shown in Figure 2, the real part of permittivity *�<sup>r</sup>* is plotted to the horizontal axis of the parameter space, while the vertical axis corresponds to the real part of permeability *μr*. A negative value of *�* (*μ*) indicates that the direction of the electric (magnetic) field induced inside the material is in the opposite direction to the inbound incident field. Noble metals at optical frequencies, for example, are materials with negative *�*, and negative *μ* and can be found in ferromagnetic media near a resonance. Waves can not propagate in material in the second and fourth quadrants, where one of the two parameters is negative and the index of refraction becomes purely imaginary. In the domain of optics, all conventional materials are confined to an extremely narrow zone around a horizontal line at *μ* = 1 in the space, as represented by the dashed line in Figure 2.

Scores of such materials are designed to manipulate EM waves, many passively, by creating an alternate propagation path. Metamaterials have been designed to redirect, not absorb or 4 Will-be-set-by-IN-TECH

The development of an efficient method for acquiring and converting conventional energy from the sensors such as solar and wind has seen an exponential growth over the last few years. The sensing development has been referred to as energy harvesting. One factor contributing to the enjoyment of such an increase has been the threat of rapid decreases projected in our global and national energy reserves based on utilization rates and trends. Such a premonition has spurred funding for research in various fields including materials or

Metamaterials have been defined by most associated scientist as materials made by man which exhibit non-natural properties and characteristics, particularly EM or electromagnetic properties not known to exist with any other materials found in nature. Regarding electromagnetism, metamaterials which exhibit propagating electromagnetic waves (both the permittivity and permeability are negative as seen in Figure 1 have seen much attention in

recent years as well as when both permittivity and permeability are very close to 1.

Fig. 1. The parameter space for *�* and *μ*. The two axes correspond to the real parts of permittivity and permeability, respectively. The dashed green line represents non-magnetic

The reaction or response of a material (as in a sensor for WSNs) to external fields is largely determined by only the two material parameters *�* and *μ*, permittivity and permeability respectively. As shown in Figure 2, the real part of permittivity *�<sup>r</sup>* is plotted to the horizontal axis of the parameter space, while the vertical axis corresponds to the real part of permeability *μr*. A negative value of *�* (*μ*) indicates that the direction of the electric (magnetic) field induced inside the material is in the opposite direction to the inbound incident field. Noble metals at optical frequencies, for example, are materials with negative *�*, and negative *μ* and can be found in ferromagnetic media near a resonance. Waves can not propagate in material in the second and fourth quadrants, where one of the two parameters is negative and the index of refraction becomes purely imaginary. In the domain of optics, all conventional materials are confined to an extremely narrow zone around a horizontal line at *μ* = 1 in the space, as

Scores of such materials are designed to manipulate EM waves, many passively, by creating an alternate propagation path. Metamaterials have been designed to redirect, not absorb or

more specifically metamaterials.

materials with *μ* = 1 Cai &shaleav (2010).

represented by the dashed line in Figure 2.

Fig. 2. Examples of a 3D optical metamaterials fabricated in a layer-by-layer manner. (a) A near-infrared NIM (Negative Index Material) with three functional layers made by EBL (Electron Beam Lithography); (b) Four layers of SRRs (Split Ring Resonators) based on EBL with patterning-and-flattening approach; (c) A NIM wedge exhibiting negative refraction for visible light made by an advanced FIB technique Cai &shaleav (2010).

reflect but to divert the energy through a desired path. It is no wonder as to the attention metamaterials have seen for energy harvesting. Research is currently being conducted to develop sensors (both photon and electron based) that extract atmospheric energy regardless of the incident angle such that no energy is reflected back out of the sensor rather, it's reflected down toward the detector. This will lead to the creation of ultra-efficient sensors for wireless networks, see (Narimanov & Kildishev, 2009; Shalaev et al., 2005) for more information on Metamaterials.

#### **3.2 Microcontroller**

The component responsible for doing the bulk of the switching and decision making for the WSN at the remote site is the processor or microcontroller. When selecting the processor for specific WSN applications, the engineer must make many considerations. These considerations include, but are not limited to, cost, power requirements, physical size, weight and speed, some of which are elaborated upon below.

Depending on the microcontroller, the power requirement could range from .25 mA to 2.5 mA per MHz for either 8 or 16 bit processors. Again, the application will determine if a processor consuming relatively high amounts of energy is acceptable or if .25 mA per MHz is needed. A common misconception is that by putting the processor in "sleep" mode, the sensor utilizes less power thus is more efficient. It has been shown that this is not always true as while in "sleep mode", sensors still maintain synchronization and memory functionalities necessary to perform expeditiously upon awakening (Hu & Cao, 2010).

In fact a more prudent approach to saving energy would include completely shutting the processor off, entirely, and ensuring the sensor can rapidly recover from a "dead" state or at the very least rapidly jump from "sleep" mode to "awake" mode. As the processor needs to synchronize native clocks and stabilize, the transition time or delay can be as long as 10 ms which is a relative eternity. Another parallel approach involves varying the speed depending on the time allotted for a specific task.

In other words, only using the minimum power required for a task at a given time by dynamically ramping up or down the power accordingly versus drawing full power for all "awake" states. This approach may benefit from an algorithm in which the speed is a function of the power. If the required task and itÕs effort expended is known before the task is given, an absolute "finish time" can be maintained without necessarily completing the task as fast

Within the TR package, a typical TR module will consist of and follow this RF path for transmission: a common attenuator for signal suppression, a common phase shifter (depending on the application. For example, phase shifter could be used to shape the transmission pattern or radiation pattern leaving a WSN (also known as beam-steering), a driver and a high power amplifier (HPA) to boost the signal amplitude for propagation from the aperture or antenna for transmission. When receiving a signal within the TR module frequency range, which varies per application, the signal passes through a limiting filter and low noise amplifier (LNA) before coursing through a common attenuator to suppress the signal's magnitude and possibly a common phase shifter (depending on the application. For example the phase shifter can be utilized as a directional finder or filter for incident signal in a WSN). Outbound and incident signals are typically discerned by a circulator at the output/input of the module. The attenuator and phase shifters are termed "common" due to the fact that these components are used for both reception and transmission. In the following, we elaborate on a few of the key components of the TR module from Figure 3.

Power Considerations for Sensor Networks 135

The attenuator is implemented to ensure the unwanted side-lobes are suppressed, sufficiently reducing the noise in the system. It also keeps the amplifiers down stream from prematurely reaching saturation and causing unwanted non-linearities. Typically this is done only for the receiver as during transmissions, it is usually desirable to propagate as much energy from the antenna as possible. Since the attenuator basically performs the exact opposite function of the amplifier, they are typically not conjoined in series unless, in some cases, it's needed for filtering purposes. Note that all the components within the TR module are frequency matched meaning they are optimized for specific frequency ranges. Due to this inherent characteristic, attenuators can be used to suppress frequency bands without distorting the fundamental waveform. This is important for the energy efficiency of the system as the modulator can maintain relative simplicity without the need to effectively recreate a waveform which would

The phase shifter allows multiple RF signals to be controlled by way of an external stimulation such that the output of the phase shifter is of the desired phase without effecting the frequency. The phase shifter may or may not be present in the TR module. It depends on whether or not the WSN calls for a beam-forming or shaping capability which can aid in power efficiency if multiple sensors are synchronized in receive and or transmission mode for power/amplitude coupling. The amplifiers (driver and high power amplifiers) boost the signal for transmission from the antenna. The level of amplification needed depends on the efficiency of the system, particularly the aperture or antenna. A poorly matched antenna or one which has a high Voltage Standing Wave Ratio (VSWR) will demand a higher amplitude or stronger signal to

The application and placement availability of WSN will greatly affect which antenna is more suitable and efficient. Most WSN antennas are omni-directional fundamentally but are shaped by various ground effects. This crucial aspect of antenna propagation has prompted many

Considering that many WSNs rely on portable energy or power sources to power sensors, the capacity and efficiency of both the power source and the WSN is crucial in the overall effectiveness of the WSN. For most of the WSN applications, when the power source drains,

researchers to develop accurate prediction models specifically for WSNs.

subsequently cost more power.

propagate to a given target.

**3.5 Power source**

a possible rather as fast as necessary. Researchers from the University of California, Irvine (Irani et al., 2007) developed an algorithm for optimizing power consumption by varying speed below:

$$\log(z, z') = \frac{\sum\_{j} \text{such that} [r\_{j'}, dj] \subseteq [z, z'] \mathbb{R}\_j}{\text{l}(z, z')} \tag{1}$$

where g(z,z') defines the intensity of the interval [z,z'], l[z,z'] defines the length of the interval, *Rj* is the required work needed to complete the job and *dj* denotes the deadline for job j. This would allow energy and speed to be spent where it's needed most creating a dynamic fluid speed variance throughout the CPU for maximum overall efficiency. One might say, 'losing a battle here and there but winning the war'.

#### **3.3 Memory**

Memory is a crucial factor in WSNs. Particularly non-volatile memory. Non- volatile memory is defined as various forms of solid state memory which doesn't need to be refreshed or powered to maintain it's information. Examples include flash, electrically erasable programmable read-only memory (PROM) read only memory (ROM), optical discs and magnetic disks (Postolache et al., 2010).

The memory component is the means at which the WSN stores the data it acquires. The speed requirement of the memory unit depends of the nature of the WSN and its intended functionality. A rather fast memory unit may be required for certain military applications where the data acquisition speed from the memory may dictate whether or not a target is detected in time for acquisition and lock. On the other hand, a relatively slow memory unit may be acceptable for soil monitoring WSN utilized by farmers. In either case the security and reliability of the memory unit is important and both require additional power demands on the WSN. To this end, researchers have been developing ways to more efficiently processing and storing the acquired data including virtual memory protocols. Virtual memory has been shown to reduce compile-time optimizations regardless of the limitations in memory on site. One approach which generates a memory layout optimizes to the memory access patterns and attributes for a given WSN. In other words, the protocol optimizes the memory map based on the application, effectively reducing overhead (Lachenmann et al., 2007).

#### **3.4 Transceiver module**

All WSN motes will possess a transceiver or TR modules as they allow the motes to communicate in WSNs. They present the capability to transmit and receive data packets, information or signals in a relatively small package. One of the main factors which allows for such a diminutive size lies in the RF architecture. Because the TR modules transmit and receive in the same RF component there is no need for a separate architecture for each transmission or reception. Thus the isolation of incident energy to transmitted energy must be great to ensure against destructive cross modulation, unwanted dispersion and various other resultant noise, all of which would inherently degrade the efficiency of the WSN either directly or indirectly. Signal loss is of particular concern in the input/output portion of the TR module and precautions must be taken to ensure signal degradation is tolerable from a minimum threshold point of view.

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

a possible rather as fast as necessary. Researchers from the University of California, Irvine (Irani et al., 2007) developed an algorithm for optimizing power consumption by varying

) = <sup>∑</sup>*<sup>j</sup> suchthat*[*rj*, *dj*] <sup>⊆</sup> [*z*, *<sup>z</sup>*�

where g(z,z') defines the intensity of the interval [z,z'], l[z,z'] defines the length of the interval, *Rj* is the required work needed to complete the job and *dj* denotes the deadline for job j. This would allow energy and speed to be spent where it's needed most creating a dynamic fluid speed variance throughout the CPU for maximum overall efficiency. One might say, 'losing a

Memory is a crucial factor in WSNs. Particularly non-volatile memory. Non- volatile memory is defined as various forms of solid state memory which doesn't need to be refreshed or powered to maintain it's information. Examples include flash, electrically erasable programmable read-only memory (PROM) read only memory (ROM), optical discs

The memory component is the means at which the WSN stores the data it acquires. The speed requirement of the memory unit depends of the nature of the WSN and its intended functionality. A rather fast memory unit may be required for certain military applications where the data acquisition speed from the memory may dictate whether or not a target is detected in time for acquisition and lock. On the other hand, a relatively slow memory unit may be acceptable for soil monitoring WSN utilized by farmers. In either case the security and reliability of the memory unit is important and both require additional power demands on the WSN. To this end, researchers have been developing ways to more efficiently processing and storing the acquired data including virtual memory protocols. Virtual memory has been shown to reduce compile-time optimizations regardless of the limitations in memory on site. One approach which generates a memory layout optimizes to the memory access patterns and attributes for a given WSN. In other words, the protocol optimizes the memory map based on

All WSN motes will possess a transceiver or TR modules as they allow the motes to communicate in WSNs. They present the capability to transmit and receive data packets, information or signals in a relatively small package. One of the main factors which allows for such a diminutive size lies in the RF architecture. Because the TR modules transmit and receive in the same RF component there is no need for a separate architecture for each transmission or reception. Thus the isolation of incident energy to transmitted energy must be great to ensure against destructive cross modulation, unwanted dispersion and various other resultant noise, all of which would inherently degrade the efficiency of the WSN either directly or indirectly. Signal loss is of particular concern in the input/output portion of the TR module and precautions must be taken to ensure signal degradation is tolerable from a

the application, effectively reducing overhead (Lachenmann et al., 2007).

]*Rj*

*<sup>l</sup>*(*z*, *<sup>z</sup>*�) (1)

*g*(*z*, *z*�

battle here and there but winning the war'.

and magnetic disks (Postolache et al., 2010).

speed below:

**3.3 Memory**

**3.4 Transceiver module**

minimum threshold point of view.

Within the TR package, a typical TR module will consist of and follow this RF path for transmission: a common attenuator for signal suppression, a common phase shifter (depending on the application. For example, phase shifter could be used to shape the transmission pattern or radiation pattern leaving a WSN (also known as beam-steering), a driver and a high power amplifier (HPA) to boost the signal amplitude for propagation from the aperture or antenna for transmission. When receiving a signal within the TR module frequency range, which varies per application, the signal passes through a limiting filter and low noise amplifier (LNA) before coursing through a common attenuator to suppress the signal's magnitude and possibly a common phase shifter (depending on the application. For example the phase shifter can be utilized as a directional finder or filter for incident signal in a WSN). Outbound and incident signals are typically discerned by a circulator at the output/input of the module. The attenuator and phase shifters are termed "common" due to the fact that these components are used for both reception and transmission. In the following, we elaborate on a few of the key components of the TR module from Figure 3.

The attenuator is implemented to ensure the unwanted side-lobes are suppressed, sufficiently reducing the noise in the system. It also keeps the amplifiers down stream from prematurely reaching saturation and causing unwanted non-linearities. Typically this is done only for the receiver as during transmissions, it is usually desirable to propagate as much energy from the antenna as possible. Since the attenuator basically performs the exact opposite function of the amplifier, they are typically not conjoined in series unless, in some cases, it's needed for filtering purposes. Note that all the components within the TR module are frequency matched meaning they are optimized for specific frequency ranges. Due to this inherent characteristic, attenuators can be used to suppress frequency bands without distorting the fundamental waveform. This is important for the energy efficiency of the system as the modulator can maintain relative simplicity without the need to effectively recreate a waveform which would subsequently cost more power.

The phase shifter allows multiple RF signals to be controlled by way of an external stimulation such that the output of the phase shifter is of the desired phase without effecting the frequency. The phase shifter may or may not be present in the TR module. It depends on whether or not the WSN calls for a beam-forming or shaping capability which can aid in power efficiency if multiple sensors are synchronized in receive and or transmission mode for power/amplitude coupling. The amplifiers (driver and high power amplifiers) boost the signal for transmission from the antenna. The level of amplification needed depends on the efficiency of the system, particularly the aperture or antenna. A poorly matched antenna or one which has a high Voltage Standing Wave Ratio (VSWR) will demand a higher amplitude or stronger signal to propagate to a given target.

The application and placement availability of WSN will greatly affect which antenna is more suitable and efficient. Most WSN antennas are omni-directional fundamentally but are shaped by various ground effects. This crucial aspect of antenna propagation has prompted many researchers to develop accurate prediction models specifically for WSNs.

#### **3.5 Power source**

Considering that many WSNs rely on portable energy or power sources to power sensors, the capacity and efficiency of both the power source and the WSN is crucial in the overall effectiveness of the WSN. For most of the WSN applications, when the power source drains,

**4. The WSN layered protocol stack**

results in power being consumed.

aggregation, and compression.

to power consumption.

**4.1.1 Information fusion**

**4.1 Application layer**

The WSN layered protocol stack consists of the Application layer, the Transport layer, the Data link layer and the Physical layer. This section will cover the role of each layer and study its power consumption. The section will survey the current literature and analyze it with respect

Power Considerations for Sensor Networks 137

The application layer is in charge of collecting and processing sensor readings (including the use of data aggregation), performing time synchronization, implementing a security protocol (as needed) etc... Each one of these tasks uses one or more hardware modules and each task

Traditionally, sensor motes were designed to perform very little to no processing. They would sense the environment and send the sensing data to the base station for processing. This resulted in large amounts of packets being sent from the motes to the base station. In addition, in several sensor network applications, the motes are exposed to conditions (such as very high/low temperatures, pressure and noise) that might sabotage their measurements. It was then proposed to use information fusion (also referred to as data aggregation) techniques at the motes in order to decrease the network traffic, save energy, remove outlier data, make predictions about future measurements and in general obtain better information quality by combining data from multiple sources. Data aggregation requires some amount of processing to be carried out at the motes. Data fusion can be used at different layers of the WSN protocol stack. For example, it can be used at the application layer to process sensor readings as well as at the network layer to consolidate routing information. In the following, we survey and

analyze the work that has been done on data aggregation and information fusion.

Information fusion can be categorized into three classes. Complementary, redundant, and cooperative. This classification depends on the particular application and the relationship between the motes that gather the data. In the case of complementary information, sources gather different types of data and information fusion is applied to obtain a more complete picture from data. In the case of redundant information, one or more sources gather the same type of data and information fusion is used to discard the outlier measurements and filter the data for accuracy, reliability and confidence. In cooperative information fusion, two sources gather information that is fused to produce information that better represents the reality. Information fusion is performed for different purposes. In the following, we present a classification of data fusion algorithms based on the purpose of the information fusion.

Information fusion techniques could be either centralized or distributed. Centralized techniques have a single point that controls the fusion process but are simple to implement. However, all the sensor motes send their data to the central point, which overwhelms the central data point and floods the network with messages. Distributed techniques on the other hand are more complex to implement but are more energy efficient because the information is exchanged locally, which reduces the number of messages exchanged in the network. Several methods have been proposed for information fusion including inference, estimation,

Fig. 3. Transceiver

the WSN is inoperable. For many applications various protocols for maximizing the lifetime of the WSN are adequate while many other applications require WSNs to remain in remote areas for several months or years without opportunities for manual power replenishment. Many research centers have developed models to efficiently harvest energy for power as for sensing previously mentioned. A WSN which can obtain its power requirements from its surrounding environment essentially has an infinite lifetime. Various approaches from mechanical vibrational energy harvesting to photon collection schemes are being considered in an effort to self-generate power needs.

Fig. 4. A low power wireless sensor node system powered from energy scavengers or harvesters and a battery. Guilar et al. (2006).

Figure 4 is a schematic of a low power WSN system that uses energy scavengers. In Figure 4, the energy sources are labeled Vsolar, Vvibe and Vbat for the solar, mechanical vibration and battery, respectively. A mutiplexer switches between the unregulated energy sources. ADC denotes the Analog to digital converter, DSP denotes the Digital signal processors and RF denotes Radio frequency.
