Modeling Multi-Point Transport Protocol in P2P Networks (1003.5413v1)

Abstract: Traditional end-to-end congestion control mechanisms assume data transferring happens between each pair user. In contrast, in a P2P network, many peers may locally keep a copy of a specific data object. If the path between a pair of peers is congested, the requesting peer who wants to download data will switch to another peer in its neighbor peer list to fetch the data instead of decreasing the download rate from the current peer. Thus, it is critical to study the performance in multi-point-to-multi-point (M2M) transport protocol in a P2P network. In this paper, we build a mathematical model for identifying the key parameters for the M2M transport protocol and also the relationships among these parameters. Finally, we conduct simulation experiments to validate our model.