A new clustering algorithm for wireless sensor networks

Wireless sensor networks have recently become an attractive research area. However, saving energy and, thus, extending the wireless sensor network lifetime entails great challenges. For this reason, clustering techniques are largely made use of. In this paper we propose a new algorithm based on the principle of spectral clustering methods. Especially, we use the K-ways spectral clustering algorithm. The main characteristic of our proposal is that it defines the optimal number of clusters and dynamically changes the election probabilities of the cluster heads based on their residual energy. Up on analyzing the impact of node density on the robustness of the proposed algorithm as well as on its energy and lifetime gains, simulation results show that the approach actually improves the lifetime of a whole network and presents more energy efficiency distribution compared to Low-Energy Adaptive Clustering Hierarch, Centralized Low-Energy Adaptive Clustering Hierarch, and Distance-Energy Cluster Structure approaches. © 2005 - 2013 JATIT & LLS. All rights reserved.