Building Trust in Wireless Sensor Networks

by
Matthew Probst

Advised by
Sneha Kasera

In wireless sensor networks not all nodes can be trusted to behave properly. One example of misbehavior is nodes that are experiencing hardware failure. Another example of misbehavior is the injection of hostile nodes into a network with the intention to disrupt communication, propagate incorrect information or collect sensitive information. Detecting such misbehaved nodes from a location external to the network becomes more difficult when algorithms for in-network data processing and aggregation are used. Wireless sensor networks can be secured most effectively against misbehaved nodes if the nodes themselves closest to the source of the problem can detect such misbehavior and react accordingly.

My research proposes the dynamic formulation of trust between nodes so as to secure a wireless sensor network against such misbehavior. I am developing and testing a trust model for a peer behavior reputation system that allows nodes to quickly detect and react to misbehaving nodes.