Node Position Estimation based on Optimal Clustering and Detection of Coverage Hole in Wireless Sensor Network using Hybrid Deep Reinforcement Learning

Abstract

Abstract Sensor nodes, generally small and low- power gadgets, are the main units of Wireless Sensor Networks (WSNs). Each node detects the changes that occur in its surroundings and sends all detected events to the sink node for analysis. However, there may be areas that aren't within the sensing range of any node since the nodes are deployed randomly. Random deployment of these sensor nodes and sometimes node failure results in coverage holes in WSNs. Time complexity increases with the size of coverage hole. Still the computational complexity is very high with various distributed methods proposed in recent times for solving coverage hole detection problem. In this paper, optimal cluster-based node position estimation and coverage hole detection in WSN using hybrid deep learning approach is proposed. First, a modified Lyapunov optimization (MLO) algorithm is used to compute the position of sensor node and it ensures edge nodes in the network. Next, we design optimal clustering using Improved Sand Cat Swarm optimization (ISCSO) algorithm to formulate effective balanced clusters which calculates coverage hole area in the network. Lastly, we developed a hybrid deep reinforcement learning (Hyb- DRL) for detecting hole shape and to estimate the hole size within clusters, among clusters and along edges. The performance of the proposed technique is evaluated using NS2 simulation tool, in which node density, node mobility, and sensing range of node are simulated.