Random sampling key revocation scheme for distributed sensor networks

Distributed key or node revocation facilitates the removal of compromised keys or nodes from a network without requiring a central authority. We propose a new revocation scheme, the random neighbourhood sampling node revocation, for distributed sensor networks. Our protocol is based on simple random sampling, a statistical method to estimate the property of the population through randomly sampling a minimal subset of the population. We exploit one of the inherent features of sensor networks: the sensor nodes are densely deployed, and there is a large overlap of the (wireless) coverage areas of any two neighbouring nodes. The revocation decision is made collectively by the neighbours of a suspicious node. However, instead of collecting the opinions of all neighbours of a suspicious node our scheme samples random subsets of the set of all its neighbours and of the node, which issued the warning. Our protocol is fully decentralized, incurs low communication cost, enables fast reaction to a detected intrusion, is false-detection tolerant and can be implemented with any pairwise key distribution scheme.