TITLE: DETECTING AND PREVENTING SELFISH NODE ATTACKS IN MANETS: A LIGHTWEIGHT APPROACH

Abstract

Mobile Ad hoc Networks (MANETs) are decentralized networks of self-organizing, mobile nodes communicating wirelessly without fixed infrastructure. Their dynamic nature, open medium, and lack of central control make them susceptible to routing attacks. A significant threat is selfish node behavior, where nodes participate in route discovery but refuse to forward data packets to conserve resources, as seen in the Ad-hoc On-demand Distance Vector (AODV) protocol. This thesis introduces a MA (Mobile Agent) based lightweight system to detect and prevent Selfish Node Type 1 (SNT1) attacks in MANETs. The proposed system employs multiple MAs that utilize a watchdog and trust value technique to monitor neighboring nodes during data transmission. By overhearing network traffic, these MAs analyze node behavior to identify selfish activity, distinguishing it from other network issues. Upon detection, the system blocks the identified selfish nodes from participating in future route discovery processes. Simulations conducted using NS-3 (Network Simulator 3) demonstrate that this proposed detection and prevention system significantly enhances network performance compared to standard AODV and existing detection mechanisms. Specifically, with 25 nodes and 6 selfish nodes, the lightweight system achieves substantially higher throughput and packet delivery ratios. While the prevention mechanism might introduce a slight increase in end-to-end delay due to the necessity of discovering new routes after isolating selfish nodes, the overall improvement in Quality of Service (QoS) by effectively mitigating data packet drops caused by selfish behaviour underscores the efficacy of this novel approach. The comparison with existing literature further validates the advancements offered by this lightweight system in addressing selfish node attacks in MANETs