Node Replication Detection in Sensor Networks

Abstract

Sensor nodes are small devices with limited capabilities. They are used for a variety of applications and can be deployed in public environments, which makes them vulnerable to physical attacks. An adversary can capture a node, read its cryptographic information, replicate these data onto multiple sensor nodes and insert them in the network.

In this thesis, we analyze two methods to detect node replicas in a sensor network, namely the randomized multicast and line-selected multicast. Randomized multicast distributes location information to a randomly chosen set of witness nodes, while line-selected multicast uses the routing topology of the network to detect collisions. We derived an analytical solution and conducted simulations to verify our predictions. The results were at par with our expectations. Both methods detect node replication with high probability, but the line-selected multicast algorithm is more efficient in terms of communication costs.