Content-Oriented Self-organization in Unstructured P2P Data Sharing Systems. An Approach to Improve Resource Discovery

摘要

The flooding-based search is a major problem in unstructured peer-to-peer file sharing systems like Gnutella or KaZaa since it results in a significant portion of Internet traffic. This article intended to demonstrate that flooding can be avoided by using a content-oriented self-organization mechanism. In contrast to connections that are usually made randomly between peers in unstructured systems, each peer set up its connections based on content interests represented by offered and downloaded files. Through the simulation, it was shown that these local activities resulted in a global content-related network topology even under highly dynamic conditions. It was demonstrated that the content-oriented topology could be used to organize focused search in order to avoid flooding.