In large scale peer-to-peer networks, it is difficult to establish the trust relationship between peers as it is less likely that repeatable interactions will occur with the same peer. So the network is vulnerable to malicious peers.

In order to settle the problem, a novel domain-based trust model is proposed in this paper. This model can be used to deal with trust relationship between the entities in peer-to-peer environment and help peer-to-peer entities make trust choice. Select a node whose performance is the optimal as a super-node in each domain of nodes. Some information of nodes in this domain is stored in super-nodes. There are a lot of this kind of domains in the whole peer-to-peer structure, and super-nodes in each domain are connected in the form of the pure peer-to-peer structure in the overall structure. This model divides the trust relationships between entities into a trust relationship in domain and the trust relationship between domains which are dealt with in a different way. This model can evaluate the trust relationship between the entities more accurately, thus can solve security issues more effectively in the peer-to-peer environment.

We present the implementing process of this model, and propose several formulae to respectively calculate the trust value in the domain and the trust value between domains. Furthmore, the simulation and analyses show that this model can get more real trust value at the small price, and the advantages of the model are simple structure and high reliability.

In conclusion, we propose a novel domain-based trust model based on the super-node network architecture. The structure of this model is very simple. The model is very easy to be accepted by users and is suitable for many kinds of peer-to-peer application environments. The trust relationships between entities are divided into trust relationship in the domain and a trust relationship between domains in the model. For the different trust relationships, we use the different calculation method to assure the real-time performance, accuracy, integrity, and calculation efficiency of the trust value. In future, we will make much more specific definition and description for this model such as updating the strategy of the trust value and so on. We will also incorporate intrusion tolerant intelligence in this model, so that nodes are able to detect false information sent by any malicious node.