SMART Service for Fault Diagnosis in Wireless Sensor Networks

In Wireless Sensor Networks (WSNs), performance and reliability depend on the fault tolerance scheme used in the system. Fault diagnosis is an important part of fault tolerance. An effective diagnosis tool helps network administrators clearly monitor, manage, and troubleshoot the performance of the network. However, the design of online fault diagnosis is crucial in WSNs since many faults can easily happen and propagate. Besides, fault diagnosis put extra burden on the sensor node and it will also consume extra resources of the sensor nodes. Thus, in order to guarantee the network quality of service, it is essential for WSNs to be able to diagnosis faults efficiently. In this paper, we propose an adaptive and efficient approach for fault diagnosis in WSN called (SMART). SMART is a layer independent fault diagnosis service for WSNs. The presented service focuses on diagnosis two types of failures that are likely to happen in WSN deployments which are the node failure due to energy depletion, and the link failure due to poor connectivity with neighbors. From the design view, SMART provides to the application many tunable parameters that make it suitable for various deployment needs: energy-robustness-detection latency tradeoffs, tolerable packet loss, reports frequency etc. Simulation results prove that SMART is resource efficient while providing satisfactory detection and diagnosis accuracy.