A hybrid fault diagnosis architecture for Wireless Sensor Networks

In remotely deployed Wireless Sensor Networks (WSN) software faults may occur due to sensor nodes process failures. Fault diagnosis remains a challenging task due to physical inaccessibility of sensor nodes, energy constraints and varying application requirements for latency and reliability. WSN once deployed resort to fault management mechanisms to ensure continued operation. While fault detection is relatively simpler, it needs intelligent augmentation from fault classification and diagnosis in order to be effective. Such diagnosis-classify regimen is either centrally managed or distributed and is triggered on need-basis or probed periodically. In this work, we present a hybrid fault diagnosis architecture that proactively probes for process failure and triggers inter-process communication aware fault diagnosis on need-basis. We also analyze reliability of proposed architecture through mathematical modeling.