Abstract:

Security is one of the most important factors impacting the practical deployments of ad hoc networks as there may be many selfish or malicious nodes in the networks. A Bayesian inference based trust model, named Bayesian-based Trust Model (BTM), is proposed to address the detection of the selfish or malicious nodes. Under the proposed BTM, nodes in the network keep monitor their neighbors through packets listening. The results are analyzed based on Bayesian inference theory, and the first-hand trust factor is derived. Based on such first-hand trust factor, the nodes derive the second-hand trust factor using another Bayesian procedure. The second-hand trust factor can then be used for selfish or malicious nodes detection. Simulation results show that the proposed BTM can detect and distinguish selfish and malicious nodes accurately, and it is more suitable for intrusion detection system and secure routing protocol than current schemes.

Key words: computer application, ad hoc network, trust model, selfish nodes, bayesian inference