Research On Trust Management Technologies In Open Peer-to-Peer Environment

As one of the core technologies in the next generation network, P2P has been widely used in many field of our daily life. At the same time, P2P network has many security problems which may result in severe loss for both government and single person. To keep the security of P2P network, the most important solution is to build trust relationship among clients, trust management technology is just the basic technology to solve trust security problems.This work makes a comparative deep research on trust evaluation and the security issues of reputation systems, including analysis for trust factors, building trust model for mobile P2P environments, and defending against collusive fraud attack for reputation systems. The main results are as follows:(1) On the research of trust factors, we find trust is bidirectional, and propose a trust model baesd on bi-evaluation. In this model, the trustworthiness of a peer is divided into service and request trustworthiness. Service trustworthiness shows the resources reliability of providers, and request trustworthiness is used to deal with requests from consumers, which can keep away malicious consumers and encourage selfish peers to share resources. A generic method for evaluating service and request trustworthiness is described. Furthermore, the implementation algorithms of such model are also depicted.(2) Since the current trust models regarding wire networks neglect the high dynamics of networks, they can not be applied to mobile P2P networks. Based on packet-forwarding, we propose a distributed trust model for mobile P2P networks. Through a distributed storage scheme, such model can make the ratings data distribute in mobile P2P networks. To reduce the computal overhead of mobile devices, a bi-feedback scheme is designed to improve the reliability of rating data. Hereby, a lightweight method for quantifying trust value and rating credible score is put forward.(3) Reputation systems are playing critical roles in P2P networks as a method to identify and suppress malicious peers. However, reputation systems are vulnerable to abuses by false ratings, and thus offering opportunities for malicious peers. By ananlyzing collusive fraud attacks and the vulnerabilities of reputation systems, we find that organized collusive clique is more harmful than loose clique. The reputation of malicious peers can be unfairly increased by launching GoodRep attack clique. Under the cover of high reputation, malicious peers can masquerade as trusted ones and violate P2P networks arbitrarily. Therefore, RatingGuard is proposed to secure reputation systems against GoodRep attack, which included a mathematical description and a modularized implementation. To identify GoodRep cliques, RatingGuard measures rating similarity of raters'activities, and thus detecting abnormalities among them.(4) Since ratings are unprotected in the transmission, malicious peers can disguise as routers to hijack and modify ratings. Based on this, we report the discovery of a novel attack, called RepHi (Reputation Hijack), against P2P reputation systems. Compared with the known attacks, such as fake raw ratings, the RepHi attackers require less effort to achieve the similar goals. We conduct a detailed investigation on RepHi, including attack model and the restricted factors of successful attack. Finally, we compare the possible defense measures and propose the defense scheme of RepHi based on the IBC signature scheme. Meanwhile, we analyze the performance of the defense scheme.