Fast Freenet: Improving Freenet Performance by Preferential Partition Routing and File Mesh Propagation

Abstract

The Freenet Peer-to-Peer network is doing a good job in providing anonymity to the users. But the performance of the network in terms of download speed and request hit ratio is not that good. We propose two modifications to Freenet in order to improve the download speed and request hit ratio for all participants. To improve download speed we propose Preferential Partition Routing, where nodes are grouped according to bandwidth and slow nodes are discriminated when routing. For improvements in request hit ratio we propose File Mesh propagation where each node sends fuzzy information about what documents it posesses to its neigbors. To verify our proposals we simulate the Freenet network and the bandwidth restrictions present between nodes as well as using observed distributions for user actions to show how it affects the network. Our results show an improvement of the request hit ratio by over 30 times and an increase of the average download speed with six times, compared to regular Freenet routing.