Audio On Demand Based On Combination Of P2P And C/S

Traditional streaming media on demand is based on the Client/Server model, butthere is a server performance bottleneck and single point of failure. As a new datadistribution mode, P2P technology completely breaks the dependence on server, so thatdata transferred directly between clients has been achieved. Using the P2P technology instreaming media service provides a new idea in solving server’s bandwidth, server’scapacity limit and other issues which make the technology of media on demand developrapidly. And this paper is a research in the scope of P2P Audio-on-Demand.Algorithm in resource downloading is an important factor that affects theperformance of streaming media on demand. And most researches in P2P streamingdownloading algorithm focus on the field of Video-on-Demand currently, whichimplicitly assumes that the researches are also applied to Audio-on-Demand. In view ofdifference in length of file, encoding bit rate and users’ action between audio stream andvideo stream, the two kinds of media stream have different requirement on on-demandperformance. The different requirement is mainly manifested in start-up delay, pieceselection algorithm and program pre-download.In this paper, we put forward a design of Audio-on-Demand based on combinationP2P and C/S. The main contents are as follows:Firstly，a download algorithm based on combination of P2P and C/S which isapplied to the Audio-on-Demand is designed:1. Against the start-up delay of the Audio-on-Demand, the paper puts forward adesign of start-up delay strategy, which is analyzed in two aspects. And the two aspectsare the download of initial data and the size of audio play buffer.2. In order to release server pressure which caused by piece selection algorithm, weraise a server load optimizing algorithm including the PSA(partition selection algorithm)and P2P-S download method. PSA is used for piece selection, and it aims at keeping thebalance of download speed and order of playing. P2P-S download strategy is used forensuring the fluency of audio playback, which optimizes the performance of Audio-on-Demand by applying server downloading threads. 3. We propose a program pre-download algorithm according to the characteristicsof audio media streaming and users’ listening habits. In this way the users can downloadaudio files in advance while listening to the current audio, so that they can enjoy thenext one at the end of the current audio seamlessly.Finally, according to the download algorithm based on combination of P2P and C/S,we design an Audio-on-Demand client software based on program list to realize playingaudio in real-time and the client software is named Audio-on-Demand Player.