A Distributed Approach for Hidden Wormhole Detection with Neighborhood Information

Ad hoc networks are promising but are vulnerable to selfish and malicious attacks. One kind of malicious attack, hidden wormhole attacks, can be mounted easily and be immune to cryptographic techniques. Wormholes distort network topology, and degrade the performance of applications such as localization and data collection. A wormhole attack is one of the most severe threats to an ad hoc network. Unfortunately, most state-of-the-art wormhole detection algorithms are not practicable. We observe and prove that, nodes attacked by the same wormhole are either 1-hopneighbors or 2-hop neighbors, and with a high probability, there are 3 nodes, which are non-1-hop neighbors, in the intersection of the two neighbor. However, such phenomena will not be present in normal topology. Thus a novel distributed algorithm is designed for wormhole detection and isolation with polynomial complexity. The detection probability is discussed. Simulation results show that the algorithm performs well regarding detection probability, as well as network overhead, false node alarms and miss detection.