Exploration of Malicious Attack and its Preventive Measure in MANET through Neural Network

Abstract

A mobile ad hoc network (MANET) is a dynamic, infrastructure-free network where wireless nodes spontaneously connect and communicate with each other. However, the absence of a fixed infrastructure makes MANETs susceptible to various security threats, including black hole attacks. In a black hole attack, malicious nodes disrupt network operation by intercepting and withholding data packets. Analyzing and mitigating such attacks is crucial for maintaining network performance and security. This thesis focuses on the analysis and understanding of black hole attacks in MANETs, emphasizing their impact on network integrity and performance. It highlights the challenges posed by coordinated attacks, where multiple malicious nodes collaborate to maximize disruption while making detection and analysis more complex. The experimental setup for individual attack analysis is carried out using the NS2 platform on a Linux operating system. The research is divided into three phases, the first phase establishes a baseline with no attacks, the second introduces a black hole attack within the same network configuration, and the third phase employs predictive algorithms to identify and mitigate the attack. A research study focused on utilizing Neural Networks and AODV (Ad hoc On-Demand Distance Vector) routing protocol for safeguarding Mobile Ad Hoc Networks (MANETs) against malicious attacks. The study highlights key vulnerabilities of MANETs due to their decentralized nature and dynamic topology, emphasizing threats like blackhole attacks, wormhole attacks, and denial of service attacks. AODV is identified as a valuable routing protocol to counteract these threats by dynamically managing network routes, while Neural Networks offer the ability to detect and prevent attacks through anomaly detection. newline