Particle Swarm Optimisers for Cluster formation in Wireless Sensor Networks

We describe the results of a performance evaluation of four extensions of Particle Swarm Optimisation (PSO) to reduce energy consumption in wireless sensor networks. Communication distances are an important factor to be reduced in sensor networks. By using clustering in a sensor network we can reduce the total communication distance, thus increasing the life of a network. We adopt a distance based clustering criterion for sensor network optimisation. From PSO perspective, we study the suitability of four different PSO algorithms for our application and propose modifications. An important modification proposed is to use a boundary checking routine for re-initialisation of a particle which moves outside the set boundary.