DHT Based Content Service Technology Research

P2P is one of the fastest growing and most popular Internet technologies. Among various P2P applications, content service is always a most important one. Generally speaking, construction of content service model, content storage, content publication, content searching and content download are the base blocks of content service. How to implement efficient content service in a complex network environment which contains huge-amount nodes is a challenging research area and of great importance.The research in this thesis focuses on content service in P2P network. It analyzes the particularity of content service in P2P network and current research status on content service. In order to provide an efficient and extensible content service based on natural semantic in P2P network, we did much research on the key technologies and make contributions to the following areas:1) An open content service model is proposed based on DHT and the semantic of resource content. This model is an OpenDHT structure, using two-steps content publication. With this design, it inherits the redundancy, fault toleration, extensibility and self organization from DHT while makes full use of the advantages provided by semantic-based searching algorithm. Besides, this model is easy for use and adapted to unbalanced network.2) A content storage algorithm adapted to hybrid DHT is proposed. This algorithm is based on verification of communication nodes and distributed resource storage, which fulfill the requirement of data integrality and system security with reasonable cost.3) A content publication algorithm which publishes index and resource itself into two steps is proposed. With this algorithm, on the one hand, node can flexibly make a resource publication decision based on its own capability and network connection; on the other hand, the using of index efficiently restrain network redundancy and prepare well for quality searching.4) A content searching algorithm based on DHT and semantic vector projection computing is proposed. This algorithm takes into consideration relation between words, improving traditional keywords extraction algorithm and projection computing method with keyword vectors. In that way, not only the keyword vectors can better present the content semantic of resources, but also the computing and communication load reduce. 5) In order to provide a better content service in a hybrid P2P network, we also take into consideration content download algorithm, user incentive mechanism and security mechanism, which guarantee the integrality of our system.