2.2 Wireless Sensor Network Architecture
2.2.3 Resource Allocation for Wireless Sensor Network
Compared to fixed sensor networks in which only the individual sensing system is deployed at the place of interest, wireless sensor nodes in addition to the sensing system also integrate all other required resources necessary for complete opera- tions. These include energy sources, radio communication transceivers, external memory and processor units. With all these components considered the wireless node resources are then constrained by the physical size and cost which are de-
sired to be kept as minimal as possible [Shahzad,2014]. This subsection presents
a discussion on the main constraints of WSNs. The challenges include energy supply, available memory and processing capability of the sensor nodes.
2.2.3.1 Energy Sources in WSNs
The advancement of the development of Micro-Electromechanical Sensing (MEMS) devices supported by vast reduction in size and power consumption of CMOS cir- cuitry has led to the production of ultra-low-cost sensing devices desired in a large number of applications. The low-cost and small-size requirements of the wireless sensors result into constraints of the type of energy sources and capacity that can be integrated into a sensor node. The three main supply categories include battery, solar sources and direct distribution.
Electromechanical energy storage in batteries is the predominant means of power supply to wireless devices today. The main forms of battery storage available for WSNs include Macro-Scale Batteries, which are considered very stable and versatile in all the small power sources hence preferred for most WSN applications. Next is a group of very small Micro-Scale Batteries. These give a small power output due to surface area limitations hence not recommended over the later. While power sources are fundamentally energy reservoirs, power-scavenging sources are characterized by their power density instead of energy density. In comparison to the energy density of the power reservoirs with usable power that is depen- dent upon the time over which they can operate, the energy provided by power scavenging sources only depend on how long the source stays in operation. Energy scavenging and harvesting differ slightly depending upon the sources used.
In [Steingart, 2009], energy scavenging is used to reference environments where
the ambient sources are unknown or highly irregular while in situations where the sources are well characterised and maintained, energy harvesting is used. In this category, a number of energy harvesting technologies have been used in WSNs successfully. These include the widely used solar cells both for recharging battery and capacitors used to power WSN. Successful studies on solar cell usage from
simple to complex systems were presented in [Warneke et al., 2002], [Roundy
et al., 2003], [Jiang et al., 2005b], [Dutta et al., 2006], [Gonz´alez et al., 2012],
[Mukter et al., 2014]. Vibration methods that include piezoelectric materials,
inductive and capacitive systems have also been used successfully [Roundy and Wright, 2004], [Lin et al., 2013], [Lee et al., 2014], [Lefeuvre et al., 2006] and
[Mitcheson et al.,2008]. Thermoelectric energy generators (TEG) which produce electrical energy directly from heat have also been used successfully to power
Wireless sensor nodes [Knight et al., 2008], [Knight and Collins, 2009], [Cassidy
and Scruggs, 2013]. TEGs are deployed at locations with steep temperature dif-
ferences in close proximity, such as interface between air and water, and, air and soil. They can also be employed on human or animal bodies to utilise the tem-
perature difference between the bodies and the air [Dewan et al.,2014]. Detailed
chronology of work done to improve these methods were presented in [Roundy et al., 2004], [Steingart, 2009]and [Dewan et al., 2014].
From the forgoing discussions, research work is ongoing in order to improve the efficiency of the existing energy sources used with wireless sensors in various ap- plication environments. These will further enhance availability and performance hence performability studies are recommended for best QoS in WSNs.
2.2.3.2 Memory in WSN
Advance developments towards low-cost and small-size sensor nodes have also affected the limits of the storage resources. Wireless nodes are equipped with memories having low capacities that limit their data storage ability. On the other hand, the choice of memory type may also be dictated by the level of power consumed when accessing data or just for maintaining data in memory. A detailed
discussion on the choice of memory was presented in [Shahzad, 2014]. From the
data sheets, a wide range of memory categories are used in various wireless sensor node platforms.
Though external flash and removable SD cards are used, memory available to individual sensor nodes is still restricted, limiting the amount of data that can be stored and/or processed at a given time. This calls for a good memory man- agement scheme that can cope with the increasing traffic demands while at the same time offering desired QoS.
2.2.3.3 Micro-Controllers in WSN
In order to achieve energy efficiency required for operations, sensor nodes are developed with low-end processors (Micro-controllers) that enable low-cost and low-power consumption. The performance of the processor is controlled by an on-chip limited memory with operating frequencies between 1-to-4 MHz, thereby reducing computation capabilities of the processor [Gabriel and Popa], [Shahzad, 2014]. However, many controllers are static and able to provide frequencies be- tween 0-to-8 MHz. Additionally, integrated on to the controller chip are Analogue to Digital Converters (ADCs) and digital Input/Output devices for providing con- nectivity to desired external devices like additional memories and sensors. The choice of a good micro-controller, therefore, considers a compromise on a number of parameters including:
1. Voltage requirements and Power consumption, which determine the amount of voltage supply needed to run the controller. This ranges between 2.7V and 3.3V in the majority of low-voltage micro-controllers. Low-power consuming controllers are most preferred for efficient energy saving. For
example, from [Texas,2003] it is noted that power consumed in sleep mode
varies between controllers with a significant range of 1µA - 50µA. This implies the need for making a better choice that may be a compromise with other factors.
2. Wakeup time is significant reducing delays that may result from frequent sleep given the controllers are expected to enter sleep mode most times of operation. A quick wake-up time will ensure the processor is not kept awake
even during short periods of inactivity [Hill,2003].
3. Peripheral support are used to provide interface between the controller and external devices through input/output port. These include digital sen- sors, transceivers, external memories and ADCs where analogue signals require conversion to digital signals. A variety of peripheral devices exists, but an excellent choice is necessary for good performance and better energy conservation.