Efficiently detecting wormhole attacks in sensor networks by information potential

Abstract

Several protocols have been proposed to defend against wormhole attacks in sensor networks by adopting cryptographic methods, packet leashes, directional antennas, or visualization, etc. In this paper, we propose a local energy-efficient method to detect wormhole attacks. This method can be applied in sensor networks with metric-based routing protocols and is independent of data aggregation. We define two parameters, said to be potential and gradient respectively. The former is to describe the intensity of attraction of a node to information flow in sensor networks, and the later is to describe the increase rate of potential from children node to parent node. A node with overmuch potential and gradient over its children will be suspect by their sons, and then these sons will tell whether their parent are malicious with the help of the algorithm proposed to estimate neighborhood potential configuration. A malicious node will be found out because its potential is a local maximum in its neighborhood. We simulate the procedure of potential computing and analyze the performance of the method.