A friendly neighbor selection algorithm based on bandwidth estimation
Shen Hai-cen · Journal of Chongqing University of Posts and Telecommunications · 2011
To improve the download efficiency of a BitTorrent system without the tracker,this paper presents a friendly neighbor selection algorithm based on bandwidth estimation.The algorithm consists of two phases.In the first phrase,the bandwidths of peers are estimated based on their upload traffic and time interval.In the second phrase,the friendly neighbors with approximate bandwidth are selected to upload preferentially according to the principle of bandwidth matching.The simulation results show that the peers' upload bandwidth was accurately estimated,which can be used for friendly neighbor selection,and the average download time in a BitTorrent system is stably reduced and the average download time of peers with high and medium bandwidth is dramatically decreased by the algorithm without the intervention of the tracker.