Learning from Routing Information for Detecting Routing Misbehavior in Ad Hoc Networks

Owing to ad hoc wireless networks' properties, the implementation of complex security systems with higher computing resources seems troublesome in most situations. Therefore, the usage of anomaly or intrusion detection systems has attracted considerable attention. The detection systems are implemented either as host-based, run by each node; or as cluster/network-based, run by cluster head. These two implementations exhibit benefits and drawbacks, such as when cluster-based is used alone, it faces maintaining protection when nodes delay to elect or replace a cluster head. Despite different heuristic approaches that have been proposed, there is still room for improvement. This work proposes a detection system that can run either as host- or as cluster-based to detect routing misbehavior attacks. The detection runs on a dataset built using the proposed routing-information-sharing algorithms. The detection system learns from shared routing information and uses supervised learning, when previous network status or an exploratory network is available, to train the model, or it uses unsupervised learning. The testbed is extended to evaluate the effects of mobility and network size. The simulation results show promising performance even against limiting factors.