Efficient Sensor Fault Diagnosis in Wireless Sensor Networks

Sensors in wireless sensor networks are prone to error due to their simple structure and the harsh deployment environments. Because wireless sensor networks have limited energy resources and consist of large number of sensors, this thesis proposes two frameworks of efficient sensor fault diagnosis algorithms to ensure network functionality. The first is a distributive model-based fault diagnosis framework. Different algorithms are designed under this framework for detecting and identifying spike and non-linearity faults without the use of reference sensors. These algorithms fill the gap between existing centralized model-based and distributed model-free frameworks and have benefits of being scalable, power efficient and highly accurate. In the second framework, Group Testing based algorithms are proposed for situations where the faulty sensors are rare. This study proposes both non-adaptive and adaptive Group Testing methods which evaluate sensors collectively to reduce the required number of tests. Algorithms of both frameworks are evaluated by simulated and real faulty sensor data. Results show that the distributed algorithms are able to achieve higher than 85% detection rate and ~1% false alarm under typical faulty signals. The Group Testing based algorithms reduce the required number of tests significantly while achieving similar accuracy as the traditional fault detection methods.