Face the Enemy: Attack Detection for Planar Graph Routing

Abstract

Localized geographic forwarding on planar graphs has been proposed for making routing scalable and efficient in large-scale wireless multihop networks. This approach, which is also known as face routing or perimeter routing, scales almost independently of the network size and is, hence, a promising scheme for the given class of networks. Existing work on this kind of routing focuses on the forwarding mechanisms, but does not yet consider security issues. In this paper, we analyze the specific security vulnerabilities of face routing and present several concepts that exploit its unique characteristics in order to detect attacks on the forwarding process. Our schemes are based on the same abstraction as face routing, namely the so-called faces, which are the polygons formed by the edges of the planar graph. They are defined as own entities in order to abstract from the underlying wireless multihop network. By operating on face abstraction level, the detection responsibility is shared among all the nodes that form the face, which increases the complexity of attacks, since all nodes inside one face need to collude for circumventing our security mechanisms. We implement the proposed mechanisms and perform a simulation study to analyze the effectiveness of our approach. The results show that our schemes sustain a true positive rate close to 100% and can provide improvements in the packet loss ratio of above 90% compared to security mechanisms defined at lower abstraction levels. Exploiting the face abstraction for providing security could be potentially used not only for face routing, but also for general geographic routing schemes in future work.