Research And Implementation Of Neighbor Assignment And Resource Recommendation Algorithm In Distributed CDN-P2P System

CDN-P2P technology has improved the expansion capability and content distribution capabilities of the system,but there are still some problems in the resource location.This paper studies from neighbor node allocation,system resource location and resource recommendation these three aspects,which mainly aims at the problem of resource location in CDN-P2P system:(1)In order to obtain a better performance of resource location algorithm and a higher resource recommendation quality in CDN-P2P system,we optimize the topology of the underlying P2P network in CDN-P2P system.The main job is to assign users with similar interests to be neighbor nodes,gather those nodes together.For this purpose,this paper proposes a neighbor allocation algorithm which can gather users'interest.Using the user's resource vector data and scoring data to calculate the interest similarity between users,allocate the nodes have high interest similarity to be neighbor nodes.Make the users get resource from neighbor nodes or nodes in P2P network,and reduce the load of the edge server.(2)On the basis of the first study point,this paper proposes a hierarchical resource location algorithm based on interest clustering.The first study point constructs a CDN-P2P system that gather users' interests which wants the resources that user's search as far as possible on the neighbor nodes.The hierarchical resource location algorithm based on interest clustering sends resource query requests to neighbor nodes firstly,and then sends resource query requests to super node if no hits,and then sends resource query requests to edge server if still no hits.This algorithm query resource from one level to another level in network until resource hits or TTL value is 0,then returns the result and save along the way.Experiments show that hierarchical query reduces the average search length and improves the query efficiency.(3)The resource recommendation algorithm based on user's interest changes proposed by this paper is also on the basis of the first study point.Only gather together the users have similar interests accurately,can we get a higher rate of recommendation accuracy when recommend resources to users.Because the recommended movie set is selected according to the user's predictive score sorting for resources.But the accuracy of the user's predictive score for the resources depends on the interest similarity between the users.The resource recommendation algorithm based on user's interest has weakened the impact of user for a long time before the ratings on resource recommendation accuracy that on the basis of gather the users with similar interest accurately,make recommendation accuracy increased by 3 to 5 percentage than contrast recommendation algorithm.