Novel Algorithms for Reliability Evaluation of Remotely Deployed Wireless Sensor Networks

Wireless sensor networks can be deployed in remote areas for monitoring rainforest, bio-diversity, detecting forest fire or even surveillance. In such remote monitoring applications, sensor nodes are deployed in unattended environments that make them vulnerable to different kind of failures. Hence, it is extremely important to perform a reliability analysis as a precursor to WSN deployment. This paper investigates reliability analysis and makes two contributions. First, an algorithm based on ordered binary decision diagram is proposed. Second, an algorithm based on Monte Carlo simulation is proposed to compute the reliability considering both individual component and common cause failure. There are some earlier works that focuses on either of the failure types but not both. In this work, battery model of the nodes are taken into account to have a realistic estimate of node reliability. The proposed model can be readily extended for any outdoor deployment scenario to assess reliability before actual deployment and hence explore meaningful insight regarding network design for instance, identifying critical failure sequences. The results of both algorithms for benchmark network configurations are validated for similar setting against existing literature. The results show that with more nodes network reliability gradually reaches a steady state (250 onwards) for a stable environment (low individual component failure due to transient errors) subject to moderate common cause failure probability (30%).