Wireless Sensor Networks in Nuclear Plants

A Wireless Sensor Network (WSN) consists of spatially distributed sensor nodes that cooperatively monitor physical or environmental conditions, such as radiation, temperature, sound, vibration, pressure, motion, relative humidity, or pollutants. These sensors are small in size and communicate with each other, typically over a Radio Frequency (RF) channel. Due to the small size and wireless communication, nodes may be placed very close to the observed phenomena. As these sensor nodes may be deployed and operated in hostile environments, it might be difficult or dangerous for humans to enter these areas such as in nuclear plants. Therefore, sensor nodes are expected to operate for periods ranging from days to years without any human intervention. At the same time these sensor nodes are subject to various types of faults such as sensing faults. In nuclear premises, the sensor reading plays a key role in the assessments of the system state. The indication of an abnormal state may be the result of a sensor failure rather than a system failure. Failure to identify the source of the indication of an "abnormal state" and take appropriate corrective action could result in expensive and unnecessary system shutdowns or, worse still, accidents that endanger both system and personnel. So, it is very important for a monitoring and diagnostic system to distinguish between the case where a sensor failure and not a system fault is responsible for the indication of an abnormal state. In this thesis, a system based on a WSN to monitor the radiation level around and inside a nuclear power plant is designed and investigated using the Castalia Simulator. A new approach for validating the sensor reading and detecting a new event is proposed. We use one of the standard routing techniques to create clusters with cluster head, and cluster-members. The fault detection algorithm is applied at the cluster-head, and is based on the idea that the sensor data tends to be correlated in both time and space: Existence of temporal correlation means that the observed reading at one time instant are related to the observed readings at the previous time instants while spatial correlation implies that the readings from sensor nodes that are geographically close to each other are expected to be largely correlated.By using the idea of Triple Modular Redundancy, three reference sensor nodes are used in each cluster. We use the two out of three voting mechanism to select a good sensor as a reference for checking the correctness of other sensors in the same cluster. The event detection algorithm is applied at the sink node and exploits the historical data maintained at the sink node. The simulation results show that the proposed fault detection algorithm outperforms other existing algorithms in terms of fault detection accuracy/ false alarm rate, number of received/ sent messages per node, fault detection processing time, and energy consumed