A multidimensional trust inference model for the mobile Ad-Hoc networks

Due to the absence of the central administrator and the dynamic network topology of the Mobile Ad-hoc Networks (MANETS), trust management is a major issue for designing the routing protocols of the network. In this paper, to identify the malicious nodes in the ad-hoc network, a trust inference model based on recommendations is proposed. This multidimensional model is aware of the energy, the channel and history of the interactions between the nodes. The proposed trust model evaluates trust by using two main components: the direct interaction trust (Primary trust) calculated using the Bayesian statistical method and recommendations from the neighbors (secondary trust). The metrics used for calculating the primary trust are: the received signal strength, time of the each interaction, the intimacy of nodes and the frequency of interactions. This trust inference model is implemented on top of the conventional Dynamic Source Routing (DSR) protocol. The results of the simulations affirms that the proposed model enhance effectiveness of routing in existence of malicious peers in the MANETs.