Detection of dishonest trust recommendations in mobile ad hoc networks

Indirect trust based on recommendations from peer nodes forms an important component of trust management system in MANET. It can provide a node the confidence to interact with peer nodes. However this indirect trust is vulnerable to various types of attacks from dishonest recommenders. This paper proposes a scheme that removes the dishonest recommendations before utilizing the recommended trusts. Removing the dishonest recommendations will remove impact of dishonest recommendations on the indirect trust of a node. The mechanism is based on the dissimilarity function among the recommended trust for filtering out the dishonest recommendations. The recommendations that result in the highest dissimilarity are considered as malicious recommendations. The effectiveness of the proposed scheme is evaluated against bad mouthing, ballot stuffing and random opinion attacks. The results show that the scheme can effectively filter out the dishonest recommendations.