A Comparative Study of Modified BBO Variants and Other Metaheuristics for Optimal Power Allocation in Wireless Sensor Networks - Advances in Heuristic Signal Processing and Applications - page 79

5.1 Introduction
Wireless Sensor Network (WSN) is a system of spatially distributed sensor nodes
with the abilities of sensing, computing, and communicating through wireless chan-
nels. Development of WSNs is motivated by many applications such as environment
monitoring, security, and detection of remote parameters [ 2 ].
In a distributed detection system (also called decentralized detection system)
[ 24 ], every sensor node performs some preliminary processing of data in a dis-
tributed manner and transmits a local decision to central node (called a sink or a
fusion center). In turn, the fusion center processes the received data and selects one
of a few hypotheses for the final decision-making. The main difference between
this approach and the classical centralized decision system is that the fusion center
has no access to the raw observation made at each sensor. Evidently, a distributed
sensor system is suboptimal compared to a centralized system in which the fusion
center has access to the observations from all sensors without distortion. However,
the distributed schemes offer the possibility for drastic reductions in communication
requirements and energy required to obtain an accurate estimate, at the expense of
some performance degradation [ 26 ].
Because of strict limitations on resources such as energy, bandwidth, and compu-
tational complexity, the standard problem in decentralized detection is to optimize
the performance of the system with respect to a desired performance criterion, spec-
ified as the detection error probability at the fusion center. The decision rule at the
fusion center and the local sensor decision rules need to be jointly designed to op-
timize the specified performance criterion. So the question is: How to combine the
local sensor observations, within bandwidth and power constraints, while keeping
the fusion error probability under a required threshold?
This chapter aims at developing a numerical solution for the optimal power
scheduling in WSN for correlated observations [ 4 ]. Three constrained variants of
the Biogeography-based Optimization (BBO) algorithm have been proposed to ad-
dress this issue. They are named as Constrained BBO (CBBO), CBBO-DE, which
incorporates the mutation procedure inherited from Differential Evolution (DE) [ 22 ]
to replace the BBO-based mutation, and 2-Stage-CBBO-DE where the population
is updated by applying, alternately from one iteration to the next, the BBO and DE
updating methods [ 5 ]. Constrained versions of DE, Genetic Algorithm (GA), and
Particle Swarm Optimization (PSO) algorithms are also developed in order to com-
pare the result with the three algorithms mentioned above.
The rest of this chapter is organized as follows: Sect. 5.2 provides a formulation
of the distributed detection problem [ 27 ]. The problem is described considering the
special case of binary hypothesis testing problem, where the optimal decision rule
is expressed in terms of the Likelihood Ratio (LR) statistic. In Sect. 5.3 , the optimal
power allocation problem is considered under the assumption of correlated and i.i.d.
observations. Section 5.4 briefly describes the conventional BBO algorithm and its
constrained variants designed for the problem at hand. The experimental results and
detailed performance analysis are given in Sect. 5.5 . Finally, the conclusions are
presented in Sect. 5.6 .