Coping With Misbehavior in Mobile Ad-hoc Networks

摘要: In this work, we address the question of how to enable a system operate despite presence misbehavior. Specifically, in mobile ad-hoc network, can keep network functional for normal nodes when other do not route and forward correctly? Node misbehavior due selfish or malicious reasons faulty significantly degrade performance networks. Existing approaches such as economic incentives secure routing by cryptographic means alleviate some problems, but all. For instance, still packets on bogus routes. We propose protocol called CONFIDANT (Cooperation Of Nodes — Fairness Dynamic Ad-hoc NeTworks) cope with It enables detect first-hand observation use second-hand information provided nodes. The view node has about behavior another is captured reputation system, which used classify misbehaving normal. Once detected, it isolated from network. Reputation systems can, however, be tricked spread false ratings, accusations praise. Simple solutions exclusively relying one's own direct observations have drawbacks, they make all available. To solve problem, fully distributed that effectively safe way. Our approach based modified Bayesian estimation classification procedure. our approach, each maintains rating trust cares about. ratings capture quality an actor performing forwarding tasks. From time exchanged others; using designed, only accepted if compatible current rating. are slightly information. Trust reflect whether reported first hand summaries published likely true. updated compatibility prior ratings. redemption prevent sudden exploitation good built over introducing fading. Data entirely distributed, value collection maintained others. simulation evaluate demonstrate performance. found high even up half population misbehaves. show speeds detection while keeping number positives negatives negligibly low.