摘要:Security is one of key factors which influence the development of mobile P2P systems. However, traditional security techniques cannot be applied directly to a mobile P2P network due to some of its characteristics such as heterogeneous nature of the peers, limited-range as well as unreliability of wireless links. In the paper we propose a distributed trust evaluation model, which helps the systems to operate normally with high probability. The model uses a polling protocol and seven metrics to real-time evaluate the reputation of mobile peers. The model exhibits three interesting features not seen in previous works. Firstly, it considers voting for peers from the perspectives of both trust and distrust. This appears to be the first attempt to incorporate distrust in the polling algorithm. Secondly, it credits/penalizes a peer according to its interaction behaviors, the size of interaction and the vote accuracy. This mechanism of credit and penalization is expected to deter dishonesty or misbehavior by the entities involved. Thirdly, it effectively solves the trust problem when no prior interaction history exists, an issue that has not been addressed in many models. In the end, the model is shown to be efficient and robust in the presence of attackers through simulation.