Bittella: Optimizing the download time and the search procedure on unstructured p2p networks based on Small Worlds

In this paper we propose a new protocol for unstructured and semantic-searching based p2p networks: Bittella. It forms a three level overlay network: the lowest level is the unstructured p2p network (e.g. Gnutella). The medium level is formed by clusters per content, we call the resulting structure Small World per Content structure. This level allows the utilization of Bittorrent-like download technique in unstructured p2p networks resulting in a reduction of the download time. Finally, the third level is an overlay based on common interest (the nodes form clusters based on common interests). This level is created by applying a learning algorithm called Ranking Algorithm and permits the reduction of the bandwidth used in the search procedure in the unstructured p2p networks. Therefore, Bittella improves the traditional unstructured p2p networks like Gnutella in two main aspects: it reduces the download time and it reduces the bandwidth consumption on the search procedure.