On Estimating the Topology of an Adversarial Wireless Network

Abstract

Owing to recent interest in sensor networks for military and security applications, studies of the security vulnerabilities of these networks are becoming increasingly important. In the present paper, the problem of estimating the topology of an adversarial sensor network is considered. The adversarial network is assumed to employ strong encryption, so that its transmitted packets are assumed to be unreadable by the observer. Thus, the algorithms are required to make use of the time correlations in channel uses by the adversarial network. Assuming the use of the MACA protocol, our algorithms are capable of estimating both the routes used by nodes in the adversarial sensor network, and as the identities of the nodes that are within each other's neighborhood (i.e., within radio range of each other), so that an attack could be designed for maximum effect. Results are presented which show that route estimation can be accomplished quickly using our algorithm, and that neighborhood estimation can be accomplished in a reasonable amount of time.