Using cluster analysis to detect attackers in vehicular delay tolerant networks

Abstract

Delay- and Disruption-Tolerant Networks (DTNs) are a suitable technology for many applications when the network suffers from intermittent connections and significant delays. However, the characteristics of a DTN make most traditional strategies of detecting attackers infeasible. In this paper, we propose a Misbehavior Detection System (MDS) to defend a DTN against blackhole and greyhole attackers without the need of an initial learning phase. We evaluate our method in two scenarios using different DTN routing protocols. We show that the proposed MDS has a fast reaction time and can efficiently detect evil nodes with varying drop probabilities under different scenarios yielding a high detection and low false positive rate while saving resources in the system.