AN EVALUATION OF PIECE-PICKING ALGORITHMS FOR LAYERED CONTENT IN BITTORRENT-BASED PEER-TO PEER SYSTEMS

摘要

In this paper the performance of layered piece-picking al-gorithms for Bittorrent-based peer to-peer systems is evalu-ated and compared to traditional single layer solutions. In aBittorrent-based peer-to-peer system, the piece-picking algo-rithm needs to ensure that the pieces of an audiovisual contentare received in time to ensure smooth playback of the content.For layered content provided in multiple qualities, the task ofthe algorithm becomes more complex, as it has to consider thelayer of the pieces to be fetched in addition to the deadline.Thus, the goal of the layered piece-picking algorithm is toensure that the best possible quality under the given networkconditions is received while the playback is not disturbed. Toillustrate the advantages of such a layered piece-picking algo-rithm, its performance is evaluated and compared to a singlelayer piece-picking algorithm. The results show that layeredpiece-picking algorithms can significantly improve the qual-ity in terms of neak signal-to-noise ratio.