A Peer-to-peer Network Architecture Based On The Semantic Community

P2P, one of prevalent architectures in networks, has been studied and researched for suppoting various applications, due to its autonomy and efficiency in resource utilization since 21st century. To countering for the traditional C/S architecture, P2P networks makes sufficient use of the edged resources, and constructs overlay topology in some proper order. It realizes the interconnection between resources through efficient algorithms of data search and location.As the development of research and experiments in P2P networks in the past decade, many characteristics have been shown, such as the heterogeneity between peers (including degree, number of the sharing files and etc). Besides, the principle of small world has also been introduced into the P2P networks to build some shortcut between peers holding some associations (geographically independent) so as to facilitate search.In this thesis, after a thorough review of the existing P2P architectures, we propose a new community-based P2P architecture, called CommuSearch. It has two major functional components comparing with existing architectures: 1) exploiting sufficiently the heterogeneity of peers to construct yet semantic community based on interest mimicking the real community in human society. The new overlay topology overlaps due to the overlappping of community interests and finally forms a small world network similar to the human's association relationships, and some unstructured P2P searching algorithms are refined for the new architecture, 2) and proposing a new secure trust model - CommuTrust, which is based on community credential to score and rate the reputation of peers efficiently. Besides, unique mechanisms of intra- and inter-community data transaction have been adopted based on the trust model. We further perform some exploratory work to support semantic search in our architecture. Finally, we conduct extensive simulation experiments to evaluate and valid the performance of this new architecture, including the search efficiency and security based on CommuTrust trust model.