No abstract

No abstract

No abstract

In the Bartercast reputation mechanism of the BitTorrent-based P2P client Tribler, peers compute local, subjective reputations of other peers by applying a flow-based algorithm to a locally maintained Bartercast graph with peers as nodes and bandwidth contributions as edges. We have previously shown that the computed reputations are more accurate when a peer takes the node with the highest Betweenness Centrality (BC) in its local Bartercast graph as the initial point in this algorithm rather than itself. BC is a powerful metric for identifying central nodes in complex network analysis, but its computation in large and dynamic networks is costly, and previously proposed approximation methods are only designed for static networks. In this paper, first we assess the stability of the nodes with the highest BC values in growing synthetic random and scale-free, and Bartercast graphs. Next, we evaluate three BC approximation methods proposed in the literature in terms of their ability to identify the top-most central nodes. We show that these approximations are efficient and highly accurate in scale-free and Bartercast graphs, but less so in random graphs. Finally, we integrate the three BC approximations into Bartercast, and we evaluate the quality of the reputations they yield.

Distributed reputation systems establish trust among strangers in online communities and provide incentives for users to contribute. In these systems, each user monitors the interactions of others and computes the reputations accordingly. Collecting information for computing the reputations is challenging for the users due to their vulnerability to attacks, their limited resources, and the burst of their interactions. The low cost of creating accounts in most reputation systems makes them popular to million of users, but also enables malicious users to boost their reputations by performing Sybil attacks. Furthermore, the burst of user interactions causes an information overload. To avoid the impact of malicious users and information overload, we propose EscapeLimit, a sybil attackresistant, computationally simple, and fully distributed method for information collection. EscapeLimit leverages user interactions as indicators of trust and similarity between the corresponding users, and collects relevant and trusted information by limiting the escape probability into the Sybil area. We evaluate it by emulating interaction patterns derived from synthetic and real-world networks. Our evaluation shows EscapeLimit's effectiveness in terms of its resilience to Sybil attacks, its scalability, and its ability to provide relevant information to each user.