Research On P2P System Based On Chord Lookup Method

The core idea of Peer-to-Peer (P2P) technology is that all participating nodes are equal in status. In other words, each node is both a client and a server providing services to others and meanwhile enjoying the services from others. Applied in various fields, this model allows the resources of the Internet to be widely explored and utilized.A main challenge in P2P systems is how to efficiently search and locate the data under the condition of the absence of a central-server. Current research in this area has been focused on the structured routing algorithms. As a representative scheme of P2P, the chord-based algorithm has many salient features such as simplicity, robustness, scalability and low path length.In this paper, first of all, we summarize the theoretical basis of P2P network technologies, analyze four kinds of P2P network topology structures and describe their classical models, such as Napster, Gnutella and KaZaa. We also introduce typical kinds of resource locating algorithms based on DHT in a fully distributed topology, and then describe the Chord algorithm in detail. However, the conventional chord-based method fails to consider the physical topology of the P2P network when designing the lookup strategy, which may result in tremendous delay to physical network routing. In addition, the problem of storage space is rarely considered. To overcome these problems, we conduct the following work:1. In order to address the issue of mismatch between the physical topology and logical topology, we propose an improved Chord-based method that integrates the advantages of the Ant Colony Optimization Algorithm and the bi-directional routing lookup method. Concretely, we apply the Ant Colony Optimization Algorithm in building the chord ring to match the topology between the overlay network and the physical network. Then, we develop a bi-directional routing lookup mechanism to further speed up the process of searching for local resources. The experimental results show that the proposed algorithm is more effective than the conventional chord-based method.2. In order to remedy topology mismatch and reduce space complexity of data storage, we propose an improved Chord-based algorithm that integrates the advantages of the Counting Bloom Filter technique and the topology-aware lookup method. It deploys Counting Bloom Filters to reduce the space complexity for data storage, and a topology-aware lookup mechanism to further speed up the search for local resources. The experimental results show that our improved chord-based scheme is significantly more efficient than the conventional chord-based method.3. We compare these two methods against the conventional chord-based method, and explain their advantages and disadvantages, from several aspects such as storage space complexity, the number of look-up hops and the length of routing path in the physical network.