Enhancing dynamic recommender selection using multiple rules for trust and reputation models in MANETs

Trust and reputation models are utilised by several researchers as one vital factor in the security mechanisms in MANETs to deal with selfish and misbehaving nodes and ensure packet delivery from source to destination. However, in the presence of new attacks, it is important to build a trust model to resist countermeasures related to propagation of
dishonest recommendations, and aggregation which may easily degrade the effectiveness of using trust models in a hostile environment such as MANETs. However, dealing with dishonest recommendation attacks in MANETs remains an open and challenging area of research. In this work, we propose a dynamic selection algorithm to filter out recommendations in order to achieve resistance against certain
existing attacks such as bad-mouthing and ballot-stuffing. The selection algorithm is based on three different rules: (i) majority rule based; (ii) personal experience based; and (iii) service reputation based. Recommendations are clustered, filtered, and selected based on these three rules in order to give the trust and reputation model greater robustness and accuracy over the dynamic and changeable MANET environment.