Effective Data-Centric Reputation Systems for MANETs: A Novel Evaluation Framework

Most existing reputation systems designed to cope with selfish MANET nodes rely on heuristics and are studied via simulation, with a focus on the consistency of obtained reputation metrics and their effect on network performance given specific corrective actions. We present a novel data-centric reputation system and an analytic evaluation framework where the notion of effectiveness is internal to a reputation system, reflecting a trade-off between the distinguishability of selfish and nonselfish nodes and the degree of selfish behavior. Assuming a probabilistic traffic model, permanent node identities, generalized source routing, and a fixed subset of selfish nodes we discuss the role of reputation awareness and route unawareness at selfish nodes.