The Research And Design Of A Trusted Remote Backup Model Of Peer-to-Peer Network

With the development of the information, the data has become an indispensable element in society life and the security of the data has more and more aroused people’s concern. Remote backup was regarded as an essential means to protect data, but the traditional data backup could not work perfectly when faced large disaster because of its high cost, poor scalability and fail easily to a single point. Therefore, a cheap and reliable long-distance backup project was urgently needed.Internet technology, especially the rapid development technology P2P can make full use of geographically dispersed large idle resources in the network and solve the disadvantages existed in the long-distance data backup by its low operating cost and good scalability. But the trust problem of P2P network is prominent, so the main research of this paper is to solve the trust problem which is the core to build a remote data backup model.This paper has studied the advantages and disadvantages of the existing trust models and a improved model of P2P remote data backup model was put forward based on it. Hybrid P2P network with super node model was applied to. In the calculation of the direct trust value, the influence factors on the trust value such as the transaction number, transaction value and the importance degree of file were fully considered and at the same time it can punish the malicious nodes. In the calculation of indirect trust value, the recommended node was divided into the known and unknown recommended node. The different categories of recommended node have different credibility, and node collaborative cheating could be effectively inhibited and the punishment system was introduced when the trust value was calculated. The concept of time frames was introduced when the direct trust, recommendation trust values and total trust value computation was calculated. Trust values can be divided into long-term trust value and the short-term trust, the smaller one was regarded as the trust value. The existing replica management strategy was improved to enhance the security and recoverability of the data. Besides, the relationships between the numbers and the availability of replica were given, the replica number was appropriately controlled and network flow was reduced.PeerSim simulation experiments showed that the model in this paper had better anti-aggression against simple malicious nodes, oscillation malicious nodes and gangs’ malicious nodes malicious behavior compared with the classical trust model and nodes successful interaction rate was higher.