Balanced Clustering Approach with Energy Prediction and Round-Time Adaptation in Wireless Sensor Networks

This paper proposes three techniques to improve the energy efficiency of clustering protocols in wireless sensor networks (WSNs). The first technique reduces the periodic network re-clustering, the second adapts the rounds' durations to the cluster-heads' (CHs) residual energy, while the third one ensures fair and spatially balanced distribution of CHs over the network. For the purpose of performance evaluation, these techniques are implemented on TPSO-CR, one of the state-of-the-art clustering protocols. TPSO-CR, its enhanced version with the proposed techniques (ETPSO-CR) and LEACH-C are compared. The results demonstrate that the proposed solutions remarkably reduce and balance the total energy consumed in the network. Thereby, the time to the first node death is more than doubled in some scenarios. The improvements are ranging from 25% to 76% for several network lifetime metrics and the total number of received data at the base station is increased by 30%, on average.