Credo: Thwarting collusion attacks in peer-to-peer content distribution systems

Current peer-to-peer content distribution networks do not effectively encourage users to act as seeders; this results in suboptimal behavior as many users disconnect immediately upon retrieving files. Reputation systems can incentivize users to remain active and seed files. Unfortunately, existing reputation systems are unsatisfactory – they do not accurately capture a node’s net contribution in the face of Sybil and collusion attacks. In this paper, we present Credo, a robust credit-based reputation system with a novel reputation algorithm to compute node reputation based on the set of uploads reported by peer nodes. Credo’s algorithm employs two novel techniques: credit diversity and modeling good behavior, which can limit the reputation gains of colluding adversaries effectively. Using simulations, we show that Credo is resilient against attack and provides an effective incentive, allowing nodes with more upload contributions to experience faster downloads. We have incorporated Credo in the Azureus BitTorrent client and show that overall system performance is significantly improved in the PlanetLab deployment. Additionally we show that our implementation scales well and is capable of handling systems with large numbers of nodes.