Anomaly detection using clustering for ad hoc networks -behavioral approach-

Mobile ad hoc networks (MANETs) are multi-hop wireless networks ofautonomous mobile nodes without any fixed infrastructure. In MANETs, it isdifficult to detect malicious nodes because the network topology constantly changesdue to node mobility. Intrusion detection is the means to identify the intrusivebehaviors and provide useful information to intruded systems to respond fast and toavoid or reduce damages. The anomaly detection algorithms have the advantagebecause they can detect new types of attacks (zero-day attacks).In this paper, wepresent a Intrusion Detection System clustering-based (ID-Cluster) that fits therequirement of MANET. This dissertation addresses both routing layer misbehaviorsissues, with main focuses on thwarting routing disruption attack Dynamic SourceRouting (DSR). To validate the research, a case study is presented using thesimulation with GloMoSum at different mobility levels. Simulation results show thatour proposed system can achieve desirable performance and meet the securityrequirement of MANET.