Failure prediction, detection & recovery algorithms using MCMC in tree-based network topology to improve coverage and connectivity in 3D-UW environment

Abstract

In an Underwater Wireless Sensor Networks (UW-WSN), one of the most challenging issue is the fault tolerance during data transmission. There are various constraints like lack of power, physical damage, hardware and software problem that leads to failure or blocked sensor node. Due to the faulty sensor node, it is quite difficult to communicate the information in a particular time period. This failure affects the overall network operation performance. In order to improve the coverage, connectivity and network performance, we propose a failure prediction, detection and recovery algorithm using Markov Chain Monte Carlo (MCMC) process. In the failure prediction algorithm, the failures of sensor node are identified by using error pattern of delayed messages and the threshold limit is based on time probability distribution function. In the failure detection algorithm, the faulty sensor node is detected using threshold limit and residual energy. In the recovery algorithm, the faulty sensor node is replaced by the nearest neighbouring sensor node with high energy capacity. Theoretical analysis and experimental simulation results are evaluated based on performance metrics such as Coverage Ratio, Failure Prediction, Network Lifetime and Recovery Policy. The results shows that the proposed 3-D UW-WSN system has a maximum coverage ratio of 23.07%, maximum increase of recovery sensor nodes 30%, maximum decrease of the predictive probability of failure node 11.11% and maximum increase of network lifetime 50%. The simulation results shows better performance and the proposed method is more efficient than the coverage of static and mobile sensor in 2-D UW-WSN algorithms. The proposed coverage of static and mobile sensor in 3-D UW-WSN mechanism performs to improve coverage and connectivity, reduces the predictive probability of failure nodes and increases network lifetime.