Research On Personalized Query Method On Named Data Network

In large dynamic distributed networks,the increasing number of text and multimedia and their copies or caches pose a serious challenge to data discovery and query.Processing massive distributed data intelligently is a key requirement of content-centric networks,and the data query in multicast networks is accompanied by obstacles such as uncertain sources and network congestion.It is necessary to provide an accurate and flexible scheme to solve the problem of information retrieval and matching.Nowadays,a new network protocol and architecture named Named Data Networking(NDN)can flexibly access dynamic and decentralized hierarchical categories.By extending NDN protocol,many frameworks or applications can be developed,such as data discovery,data collection and so on.Faced with these complex distributed data,discovering and collecting the required information not only needs complex and personalized requirements description,but also needs to arrange complex routing and distributed processing components and adapt to network capabilities.Based on these needs,the innovations of this thesis are mainly manifested in the following three aspects:1.To effectively discover and retrieve distributed data in distributed network environment,this thesis proposes an effective and customizable Top-k query processing framework based on NDN network.This framework provides a complete query paradigm.It forwards consumer's request Interest through multicast,asynchronously Concast processes and returns response data from different sources,controls the cycle and revives Interest and distinguishes the completeness of response.Then,routers distributed on NDN collaboratively discover and aggregate relevant Top-k data.Finally,the routers select the k response data with the highest score by ranking and returned to the consumer.The experimental results show that this method can effectively reduce network traffic,shorten the response time of queries and improve the quality of service.2.In order to provide personalized query services and adapt to network capabilities,this thesis proposes a differentiated service mechanism for data query based on NDN according to the design principles of differentiated services in existing networks.First,inspired by the MPLS protocol label idea,personalized and complex data selection requirements are transformed into multiple labels and combining them,including Interest forwarding labels and Data returning labels.Then,in the forwarding and response phases,different Interest and Data routing rules,such as the number of sending,forwarding distance and matching mode,as well as response strategies and traffic control,are determined by the labels and their combinations.Finally,according to the ability of personalized data acquisition and adapting to network,the structure and implementation process of differentiated service system based on NDN are proposed.To some extent,the mechanism meets the individual query requirements of different performance in multi-hop Information Center network.The experimental results show that this method can effectively balance network traffic and achieve the effect of differentiated services.3.In the content centric network,the number of data transmission paths,bandwidth and adaptability have an important impact on the performance,reliability and user satisfaction of big data transmission.In view of the vulnerability of single-path routing in existing networks,the high cost of multipath routing algorithm as well as the insufficiency of dynamic adaptability and service personalization,a multipath intelligent routing algorithm for data transmission based on hierarchical naming is proposed.Multipath intelligent routing based on hierarchical naming can not only improve data transmission rate,but also discover the transmission multipath between user and single data source.It can customize efficient transmission multipath for specific data categories,evaluate multipath performance,and then an efficient multipath data transmission service is constructed intelligently,adaptively and qualitatively based on the data name and user's immediate location as well as network topology.