Towards Optimized One-Step Clustering Approach in Wireless Sensor Networks

This paper introduces a nonlinear integer programming model for the clustering problem in wireless sensor networks, with a threefold contribution. First, all factors that may influence the energy consumption of clustering protocols, such as cluster-heads selection and distribution, are considered implicitly in the model. Second, an innovative fitness function that directly maximizes the WSN lifetime is proposed. Finally, a global optimum of the whole network lifespan clustering schemes is targeted. The proposed model is then solved by a particle swarm optimization meta-heuristic based algorithm. This algorithm follows a centralized one-step off-line approach, in which the clustering schemes of the whole network lifetime and their associated durations are computed at the initialization of the network operations. The one-step approach is enabled by an energy prediction mechanism that allows to reduce the costs of the periodic network re-clustering. Simulation results show clear enhancement in network lifespan and number of received data packets as compared to some state-of-the-art clustering approaches.