Asynchronous PSO for Distributed Optimization in Clustered Sensor Networks

Abstract

In this paper, a new distributed boosting technique has been proposed based on particle swarm optimization (PSO) in order to efficiently perform the regression modeling in wireless sensor networks (WSNs). The proposed algorithm learns the network regressor in two stages: (i) the clusters regressors are learned using distributed PSO, and (ii) the accuracy of the obtained models are improved through a boosting technique. The results on real dataset show that the proposed algorithm could obtain high accurate model with completely acceptable energy consumption in comparison to other distributed algorithms.