Intrusion Detection in MANETs Using Nature Inspired Algorithms Based Clustering Techniques

Abstract

Mobile Ad-hoc Networks (MANETs) represents the interconnection of heterogeneous battery powered mobile nodes with self-organizing capability in an infrastructure less environment. MANETs possesses different computational capabilities and resources availability. In specific, the distributed and dynamic characteristics of MANETs makes them more ideal and suitable for its deployment in the conditions of volatility and extremity. They are identified to be excellent candidates for application in several domains that include rescue operations, environmental monitoring, and military operations, etc. The restricted range of wireless communication and dynamic change in node mobility necessitates maximized range of cooperation among them for during the process of facilitating networking services. Thus, the mobile nodes play an indispensable role of a host and a router. However, the distributed and dynamic nature of MANETs make them vulnerable to different kinds of attacks that include Denial of Service (DoS) attack, Internet Protocol (IP) spoofing, traffic distortion, wormhole, sinkhole, blockhole, etc. The malicious node intruding into the network increases the probability of launching attack against cooperative nodes and crumble the overall performance of the complete network. MANETs on par with its counterpart wired network does not possess a ideal checkpoints like switches and routers over which Intrusion Detection System (IDS) can be deployed. In MANETs, each mobile nodes need to run its own IDS by operating in a promiscuous mode due to the absence of any centralized monitoring entity. At this juncture, it is not possible to run the IDS continuously in mobile node due to its limited battery life. In this context, design, and development of IDS schemes with energy efficiency is essential for sustaining high accuracy and detection rate during attack detection. In this thesis, four swarm intelligent optimization clustering algorithms-based IDS schemes is proposed for achieving maximized throughput, and detectio