Showing papers by "Xiaoheng Deng published in 2008"
12 Dec 2008
TL;DR: A Trust Path-Searching Algorithm based on PSO that can get the overall optimum solution in a relatively short time after many times of iteration, and the more times it iterates, the better the trust path is.
Abstract: As there are many malicious nodes spreading false information in P2P networks, it is of great importance to build a sound mechanism of trust in the P2P environments. To avoid the shortage of the existing trust model, this paper provides a Trust Path-Searching Algorithm based on PSO. In this algorithm, after initializing the particle swarm, each particle can update its speed and location according to its information, and then produce a new particle with better value. Doing that process continuously, implementing the global search of the space, finally, we can get a better overall value, which is the better trust path in the networks. The simulation results show that it can get the overall optimum solution in a relatively short time after many times of iteration, and the more times it iterates, the better the trust path is. It can be proved that to a certain extent, the algorithm can prevent the fraud.
4 citations
18 Nov 2008
TL;DR: The theoretical analyses and experimental results show that, RON suppresses forgery and collusion attacks, has the desirable feature of reputation message costs, and this model satisfies demands of accurately computing global reputation values.
Abstract: The reputation models based on distributed hash table are not suitable for unstructured P2P networks, whereas the existing reputation model for unstructured P2P networks requires a central server for assigning a set of trust management peers for any peer. We present a new reputation model for unstructured P2P networks in this paper. Differing from existing reputation models, in our model, each peer manages global reputation values of other peers in which it is interested. Local reputation values can be computed through feedback about transaction behaviors of peers, and aggregated to obtain global reputation values through the proposed reputation overlay network (RON). RON is comprised of similar interest peers, and can be evolved according to the proposed topology evolution protocol in this paper when forgery and collusion attacks occur. The theoretical analyses and experimental results show that, RON suppresses forgery and collusion attacks, has the desirable feature of reputation message costs, and this model satisfies demands of accurately computing global reputation values.
2 citations