Random-Session and K-Neighbour Based Suspected Node Analysis Approach for Cooperative Blackhole Detection in MANET

Abstract

Cooperative Blackhole is generated by two or more attackers with mutual handshaking. The nodes behave normally in reciprocal interaction and communication. In this paper, a session based suspected node evaluation model is presented to detect and prevent the cooperative blackhole attacks. A session is established for prior communication estimation as the communication begins. The parameter specific evaluation is performed over the session to identify the distrusted nodes. After this session, the continuous observation on behaviour of suspected nodes is performed throughout the communication. K-neighbour evaluation method is applied to recognize the cooperative attack behaviour of the suspected nodes. The evaluation is performed randomly on smaller separate sessions during the communication. The status update on nodes is also done based to recognize the attacker or safe node. The degree of conjunctive interaction in a flow is observed to recognize the cooperative attacker nodes. The proposed model is simulated on the multiple networks with variations in terms of the number of nodes, node speed and the number of attacker nodes. The evaluations are taken against the independent blackhole and cooperative blackhole nodes. The comparative evaluation against AODV, Probabilistic Blackhole Detection methods verified the reliability and effectiveness of the proposed model. The qualified evaluation performed on PDR and lossrate parameters verified that the investigated model has improved the PDR ratio significantly for highly infected, scattered and dense networks.