Title: Performance evaluation of main approaches for determining optimal number of clusters in wireless sensor networks
Authors: Meryem Bochra Benmahdi; Mohamed Lehsaini
Addresses: STIC Laboratory, Tlemcen University, 13000 Tlemcen, Algeria ' STIC Laboratory, Tlemcen University, 13000 Tlemcen, Algeria
Abstract: Among important issues in the current energy-efficient routing protocols for wireless sensor networks (WSNs) based on clustering approaches are how to determine the optimal number of clusters, how to generate clusters and how select cluster-heads to improve WSNs performance. These approaches aim to reduce energy consumption of the WSN nodes and to extend the network lifetime. This paper reviews and compares the performance of three clustering techniques for determining number of clusters: rule of thumb, elbow, and silhouette. Each of these techniques involves a distributed K-means approach to generate clusters. Moreover, we compare these three clustering methods with low energy adaptive clustering hierarchy (LEACH), Imp LEACH and MODLEACH in terms of network lifetime, energy consumption and the number of packets sent to the base station (BS). The results obtained indicate that rule of thumb method provides better performance compared to other clustering techniques in terms of energy consumption and network lifetime.
Keywords: clustering; K-means; low energy adaptive clustering hierarchy; LEACH; silhouette; elbow; rule of thumb; wireless sensor networks; WSNs.
DOI: 10.1504/IJAHUC.2020.106659
International Journal of Ad Hoc and Ubiquitous Computing, 2020 Vol.33 No.3, pp.184 - 195
Received: 12 Nov 2018
Accepted: 21 Jun 2019
Published online: 16 Apr 2020 *