A Variance Difference Method for Determining Optimal Number of Clusters in Wireless Sensor Networks

Abstract

Sensor networks are frequently employed to keep an eye on rapidly changing, dynamic environments. Low latency, energy efficiency, coverage difficulties, and network lifetime are seen to be the most important problems in wireless sensor networks. Cluster-based wireless sensor networks require additional study to overcome issues with energy efficiency and network lifespan. Finding the ideal number of clusters with the goal of reducing energy consumption is one of the primary challenges in cluster-based networks. The right value for k relies on the shape and size of the point distribution in a data collection, as well as the user's preferred level of clustering resolution. Additionally, if each data point is taken into account as its own cluster, increasing k without suffering any penalties diminishes the degree of accuracy in the resulting clustering until it reaches zero. Hence, Variance Difference Method (VDM) is proposed in order to determine the ideal number of clusters K and to carry out clustering in WSN. Elbow method, Silhouette method, and Gap statistic method performance is also reviewed and contrasted with that of the suggested VDM in order to demonstrate that the proposed VDM performs better than Elbow method, Silhouette method, and Gap Statistic method.