Hierarchical agglomerative clustering schemes for energy-efficiency in wireless sensor networks

Abstract

Extending the lifetime of wireless sensor networks (WSNs) while delivering the expected level of service remains a hot research topic. Clustering has been identified in the literature as one of the primary means to save communication energy. In this paper, we argue that hierarchical agglomerative clustering (HAC) provides a suitable foundation for designing highly energy efficient communication protocols for WSNs. To this end, we study a new mechanism for selecting cluster heads (CHs) based both on the physical location of the sensors and their residual energy. Furthermore, we study different patterns of communications between the CHs and the base station depending on the possible transmission ranges and the ability of the sensors to act as traffic relays. Simulation results show that our proposed clustering and communication schemes outperform well-knows existing approaches by comfortable margins. In particular, networks lifetime is increased by more than 60% compared to LEACH and HEED, and by more than 30% compared to K-means clustering.