A Technique to Identify and Substitute Faulty Nodes in Wireless Sensor Networks

In this paper, we propose a technique to identify and substitute faulty nodes to achieve fault tolerance in wireless sensor networks. The proposed technique divides the network into disjoint zones while having a master for each zone. The zone masters are used to identify faulty nodes by virtually dividing the zone into quadrants until a suspect node is found. Our fault model assumes both communication and sensing faults which are caused by a hardware failure in anode. To detect communication faults, the division process is based on calculating the throughput for each zone and comparing it to a predefined threshold. However, for sensing faults it is based on comparing the data a node senses to a predefined status and data ranges. In addition, we make use of a new technique, which was inspired by the roll forward checkpointing scheme, to activate sleeping nodes in order to validate the correctness of the suspected nodes. This is used to reconfigure the network using fault free nodes only.