A Hybrid K-Mean and Graph Metrics Algorithm for Node Sleeping Scheduling in Wireless Sensor Network (WSN)

Abstract

Wireless Sensor Networks (WSN) has proliferated in the past decade. These networks consist of massive number of battery-powered nodes distrusted over a given area. The nodes are responsible for sensing the environment and delivering the sensed data to a central point, named sink node. In order to reduce the power consumption of these nodes, sleeping/waking scheduling strategy has been proposed. In this work, a new hybrid sleeping/waking scheduling algorithm is proposed based on graph theory metrics and unsupervised K-mean machine learning algorithm. In the proposed algorithm, the sink node is responsible for calculating the metrics and clustering the nodes into three main clusters; dense, mid and light. Subsequently, the algorithm attempts to reduce the load on the nodes in light cluster in order to prolong the network lifetime. The algorithm has been simulated in 3D WSN with a clustering routing protocol. The simulation results show that the algorithm reduces the number of working sensor network nodes without affecting the network diameter. Moreover, the scheduling strategy has prolonged the network lifetime and has reduced the number of disconnected components.