Filtering Dishonest Trust Recommendations in Trust Management Systems in Mobile Ad Hoc Networks

Abstract

Trust recommendations, having a pivotal role in computation of trust and hence confidence in peer to peer (P2P) environment, if hampered, may entail in colossal attacks from dishonest recommenders such as bad mouthing, ballot stuffing, random opinion etc. Therefore, mitigation of dishonest trust recommendations is stipulated as a challenging research issue in P2P systems (esp in Mobile Ad Hoc Networks). In order to cater these challenges associated with dishonest trust recommendations, a technique named “intelligently Selection of Trust Recommendations based on Dissimilarity factor (iSTRD)” has been devised for Mobile Ad Hoc Networks.  iSTRD exploits  personal experience of an “evaluating node” in conjunction with majority vote of the recommenders. It successfully removes the recommendations of “low trustworthy recommenders” as well as dishonest recommendations of “highly trustworthy recommenders”. Efficacy of proposed approach is evident from enhanced accuracy of “recognition rate”, “false rejection” and “false acceptance”.  Moreover, experiential results depict that iSTRD has unprecedented performance compared to contemporary techniques in presence of attacks asserted.