Based On The Characteristics Of P2p Multimedia File Search System

With the development of network communication and storage technology, multimedia,because of its rich content and intuitive performace,has became the main form of information presentation.How to effectively query and access the mass multimedia shared by users,has became a hot research topic in recent years. Characteristic-based multimedia retrieval,providing intuitive and flexible approach for Human-Computer Interaction,is regarded as one of the future directions.At the same time,P2P networks,which solves the problem of server bottom-next by non-centric archecture,gives more powers of user client software and provides innovative searching approaches.However,current P2P network cannot fully support this retrieval,and is still mainly based on the file ID and keywords because of the following two reasons.On the one hand,if characteritic-based searching applies the structured P2P technologies such as M-Chord,the character descriptor must be mapping to the one-dimension liner space according to its sementic[1],which is difficult to sufficiently apply into the actual situation.On the one hand,if unstructured search technology is applied,the charicteristics of highly distributed resource locating and frequently network chumings,is bound to increase access and compare the cost of characterization.To solve this problem,this paper magerites distributed Multimedia search algorithems named SWIM(Small World Index Method)[2][3][4],which is used in static distributed database environments,into dynamic P2P environment.To avoid the performance serverely decline caused by P2P churing,this paper integrate SWIM with dynamic adaption mechanism of P2P algorithms named Kademlia[5],thus produce the combination algorithm named SWAD(SWIM-Kademlia).The improved approach has following advantages:first,it can be widely support various kinds of MPEG-7 multimedia characteristic descriptors;second,the network is organized by unstructured manners,thus providing flex query approaches;third,the local greedy query algorithm is applied instead of the classic flooding of unstructured P2P,thus saving the query time and network overhead;last,it can dynamically adapt the P2P dynamic churing environments,and keep performance under frequently node joins and leaves.