Design And Implementation Of Resource Location Strategy And Recommendation System In CDN-P2P System

With the deepening of the integration CDN and P2P technology research, CDN-P2P network content distribution in low efficiency of resource location is more and more prominent. In order to improve the network scalability and speed of response to the request content, CDN-P2P taken in unstructured P2P network topology can effectively alleviate the pressure. But with the growing number of user nodes, easy maintainability of P2P network and the efficiency of resource location are lower and lower, and the cost to find resources is increasing. Enhancing P2P network part in CDN-P2P system maintainability, reducing the time of P2P network search to find some of the resources and improving the efficiency of resource location, are currently the hot issue of CDN-P2P systems research. In this paper, according to these problems, through the analysis of the behavior characteristics of CDN and P2P users, combined with the user demand social attributes and resources attribute information, topological structure and search algorithm of CDN-P2P network has been improved, and integrate intelligent recommendation technology to the original CDN-P2P system.The main research content includes the following three points:(1)On account of the P2P network in CDN-P2P Systems, designs a P2P subnet clustering algorithm with considering user social attribute and resources demand attribute to subnet, comprehensive user social attributes and resources demand attribute clustering and subnet, can effectively reduce the average time to find the location of resources.(2)The design of resource search strategy in CDN-P2P systems is divided into CDN part and unstructured P2P network part. The CDN part based on similarity optimize neighbors selection strategy, unstructured P2P network in the establishment of the similarity of the cluster model, we design and implement a cluster-based similarity search strategy to improve the random walk through the rooms in the cluster and the cluster are taken to flooding and random walk search, you can effectively improve the efficiency of resource search.(3)In CDN-P2P prototype system, designs a recommendation system, for the problems of user data sparsity and recommendation results meeting user interest changes, we propose a collaborative filtering recommendation algorithm with comprehensive rating residual and time weight as main recommendation algorithm of the CDN-P2P system, the algorithm can effectively solve data sparsity and improve recommendation accuracy.