An Analytical Black Hole Attack Model Using a Stochastic Topology Approximation Technique for Reactive Ad-Hoc Routing Protocols

This paper presents an analytical Black Hole attack model to predict the mean packet loss of ad-hoc networks using reactive routing protocols without apriori knowledge of the actual topology configuration. Topology information is summarized as a set of prototypical hop-distance profiles that describe likely hop distance perspectives within the topology and are generated using K-means clustering. Experiments are conducted to validate the theoretical attack condition, the hop-distance profiles, and prediction performance. Results show the model prediction falls within the 95% confidence intervals of packet loss through simulation of a variety of fixed and ad-hoc topologies.