A comparative study of modified BBO variants and other metaheuristics for optimal power allocation in wireless sensor networks

Abstract

This chapter studies the performance of a wireless sensor network in the context of binary detection of a deterministic signal. The work considers a decentralized organization of spatially distributed sensor nodes, deployed close to the phenomena under monitoring. Each sensor receives a sequence of observations and transmits a summary of its information, over fading channel, to a data gathering node, called fusion center, where a global decision is made. Because of hard energy limitations, the objective is to develop optimal power allocation schemes that minimize the total power spent by the whole sensor network under a desired performance criterion, specified as the detection error probability. The fusion of binary decisions is studied in this chapter by considering two scenarios depending on whether the observations are independent and identically distributed (i.i.d.) or correlated. The present work aims at developing a numerical solution for the optimal power allocation scheme via variations of the biogeography-based optimization algorithm. The proposed algorithms have been tested for several case studies, and their performances are compared with constrained versions of the differential evolution algorithm, the genetic algorithm, and the particle swarm optimization algorithm.