The Research And Application Of Searching Algorithm And Resources Optimal Allocation Strategy In P2P Network

As the quick development of Internet and computer hardware technology, more people start to exchange information and get services from networks. By the end of 2003, there is more than 8,712,000 website and the users on the Internet is exceeding 1 billion; however, within such a distributed global information system with such amply information resources, computation power and storage capability, there has been a lot of problem such as access delay, communication errors and unbalanced load. Researchers try to solve these problems on the Internet by using Web Cache, Cluster and CDN to improve the capability and robustness of server, decrease the access delay and reduce the unnecessary communication load. But the recent date from last ten years still shows overloaded server and degraded services. In order to solve these problem, more and more designer has focused these research into one area—Peer-to-Peer networks and Peer-to-Peer computing.Most of the services on the Internet are based on C/S model. C/S based services use centered data storage and network management, which simplified the resources searching process, but on the other hand it also brings problems such as local overload, single point of failure and low extensibility. P2P model free itself by abandoning the server of C/S model, which means every point in the net could be a server or client. In the P2P networks, clients get resources and services without the help from server. So P2P is good at decreasing the accesses delay, improving the extensibility, increasing the data consistency and security.This paper deals with the core problem in the P2P networks—peer searching and discovery. According to different types of P2P networks, this paper discuss three kinds of most popular peer searching and discovery algorithm—flooding, index and dynamic hashing. Furthermore, it makes a close comparison and analyzes of these algorithms by case based studies. The object optimal allocation is one of the most important strategy that effect the extensibility and stability of P2P networks, this paper has proposed a mathematic modeling of this problem and base on that model, it compares two famous object allocation algorithm—Hill-climbing and heuristic algorithm.For the application of multimedia system in P2P networks, this paper proposes a hybrid model with both C/S and P2P model, which combines the privileges of these two models in order to ensure the extensibility of the system and the quality of service. This paper emphasizes its discussion on the realization of object optimal allocation in this model. The algorithm for this model achieve the system optimal allocation by trying to find a optimal allocation for one certain file using a mechanism much like routing table. This algorithm's analyze shows it has good computation complexity and information complexity, and can be used in various distributed computation environment.For the application based on P2P networks—P2P computing, this paper make an initial try. It builds up a P2P message distribution system, this system don't need the traditional server and all the communication are point-to-point, which means every host can distribute or subscribe to a message which he/she is interested in. The advantages of this system are powerful searching capability and extensibility; moreover, the local storage of user's data protects the privacy of P2P network users.