On the Development of Energy-Efficient Distributed Source Localization Algorithm in Wireless Sensor Networks Using Modified Swarm Intelligence

Abstract

Source localization and tracking is an important application in wireless sensor networks. Sources are localized by measuring the direction of arrival from the signal impinging on the array of sensors. Most of the existing array signals processing algorithms to estimate the source directions of arrival are centralized-based which need more communication in the network. Therefore, a distributed maximum-likelihood-based direction of arrival estimation strategy is developed by following the diffusion cooperation among the nodes in sensor network to minimize the communication overhead. Modified particle swarm optimization is proposed to optimize the multimodal maximum-likelihood function in distributed scenario. The experimental results exhibit improved performance for the distributed method over non-cooperative algorithm. Further, clustering-based approach is proposed where the nodes are clustered and then act as random arrays. Then, each cluster estimates the source direction of arrival by optimizing the maximum-likelihood function locally with cooperation of other clusters. The distributed in-clustering approach offers low communication overheads and better estimation accuracy compared to other methods.