The Research Of CDN Distribution Strategy Based On User Interest Model

With the rapid development of the Internet，especially the arrival of the3G networktime，cyber source is more and more abundant，especially in multimedia，Internet video，network television，streaming media makes the emergence of Internet bandwidth pressureincreasing. A new cyber source distribution structure----CDN emerge as the time required.CDN content distribution network，which a large extent promoted the quality of the network，greatly reduce the backbone network bandwidth pressure，and also improve the userexperience. But CDN also has the following limitations，so that it has been unable to meet therapid development of network. The first is the CDN edge buffer capacity is limited，can notstore all cyber source; Second is the result of backbone network bandwidth constraints，resource center cann`t send a copy of the resource to each CDN edge server.In view of the above problems，from the perspective of the personalized service，thisarticle do research based on a certain regional Internet user community interest. Design abased on user interest model CDN content distribution strategy，This distribution strategy canaccording to the CDN edge area to the user’s interest and decided to distribute the edge serverwhich data and the edge of the storage buffers which content. So that more users request inedge server processed.This paper uses the web log records，through clustering method to solve the user groupsinterested in. At first，to make the web log preprocessing，into standard available clusteringdata. In second，to improved k-means algorithm，improved k-means algorithm problems ofinitial arbitrary classification number K value and initial random selection of initial clustercenter. So that it can meet our user groups interested in solving demand. After clusteringresults we will carry on the keyword extraction ratio calculation，data processing，thusestablishing a user interest model. User interest model clearly marked the user groups tocertain types of resources and the degree of interest，we will transfer directly to the decisionmakers to guide CDN distribution. According to input the user interest model，the CDNcontent distribution mathematical model is improved，so that it has better performance，userrequest shorter response time.Finally，this article on the Linux server configurat CDNsim simulation environment and simulate content distribution network. Experimental verify the CDN content distributionstrategy of this paper.