Optimality analysis of locality-aware tit-for-tat-based P2P file distribution

Abstract

In the current Internet, periodic software update is essential to maintain systems secure from malicious attacks. In case of proliferated software, e.g., Operating System (OS), the distribution server for update tends to be a bottleneck, due to the concentration of users’ requests. To tackle this problem, several systems, e.g., Windows Update, have been applying the Peer-to-Peer (P2P) file distribution technique where clients called peers upload retrieved fragments of the original file, i.e., pieces, to others. However, peers may not be willing to upload pieces to others, due to their own communication overhead. BitTorrent has adopted the Tit-for-Tat (TFT) strategy in game theory, which encourages peers to exchange an equivalent number of pieces among each pair of peers. In recent years, the optimality of TFT-based P2P content distribution, i.e., file distribution and streaming, has been analyzed with the help of Integer Linear Programming (ILP). In this paper, considering the fact that the communication overhead inside a group, e.g., LAN or Autonomous System (AS), is much less than that of between groups, we model locality-aware TFT-based P2P file distribution where the TFT constraint is relaxed for intra-group communications. We further formulate the optimal piece flow determination problem as ILP in the similar way to the existing work. Through numerical results, we show that the locality-aware TFT-based P2P file distribution can achieve quasi-optimal average file download time when the upload capacity of the server is a bottleneck and the number of groups is moderate.