Machine Learning Clustering Techniques for IoT Networks: Comparative Study

Abstract

Clustering is the process of dividing a sample of data points into groups called clusters. Nodes in the same group share some commonalities, while every group differs from others. In this paper, we compare multiple Machine Learning (ML) clustering techniques. We study to what extent every considered approach compares to each other in the context of the Internet of Things (IoT) networks, the latter being based on battery-empowered devices and in which clustering is commonly used for topology management and maximizing the network durability. A comprehensive comparative study of the different clustering algorithms is pre-sented. These clustering algorithms are compared in detail based on various parameters, name-ly; the cluster number that impacts the energy consumption, the energy consumption since energy optimization is a major concern for energy-constrained wireless networks, and the net-work lifetime that indicates the network durability, the cluster stability that denotes the rate of the required re-clustering process that can be very heavy in a dynamic network.