Dealing with dishonest recommendation: The trials in reputation management court

Abstract

Reputation models play an important role in defending Ad-hoc networks, such as securing routing and data forwarding protocols, against insider attacks. However, the performance of reputation models could be easily compromised by various dishonest recommendation attacks, i.e., slandering, self-promoting and collusion. Mitigating the influence of dishonest recommendation remains an important and challenging issue in Ad-hoc networks, especially when the dishonest recommendations are in the majority. In this paper, we propose a simple, novel and effective recommendation verifying scheme (RecommVerifier) to deal with dishonest recommendation. In RecommVerifier, tackling dishonest recommendation problem is modeled as the trials in reputation management court. Then three collaborated parts including deviation detection, time verifying and proof verifying, are proposed to protect reputation model from not only individual dishonest recommendation attacks but also collective ones. The novelty of our proposal is that it does not merely depend on majority rule but introduces time verifying mechanism to reduce the false positives and false negatives caused by deviation detection. Furthermore, proof verifying mechanism, which works at the side of evaluated node, is proposed to verify whether the recommenders are honest with certainty. Experimental results show that the proposed scheme is both effective and lightweight in alleviating the influence of different types of dishonest recommendation attacks.