视频点播系统中视频分片协同存储方案研究

发布时间：2018-05-12 18:10

本文选题：网络编码 + 视频协同存储　；参考：《西安交通大学学报》2014年04期

【摘要】：为了提高视频点播系统的服务质量,减少等待延迟,改善用户体验,将视频协同存储与网络编码技术相结合,以最大化本地命中为目标,提出了一种视频分片协同存储最大化本地命中算法。利用网络编码技术,预先将视频进行分片编码。在接收到视频请求时,将编码后的视频片发给用户。在客户端接收到足够的视频片后,即可进行解码,从而得到完整的视频。该算法构造了一种资源分配有向图,将原问题转化为该图上的最小费用流问题;通过求解该最小费用流问题,在多项式时间复杂度内得到最优化方案。当视频的总容量与服务器的总容量之比较大时,该算法依然有较好的表现。实验结果表明,该算法的性能优于其他已有的算法,特别是在视频总容量与服务器总容量之比较大的情况下,性能有10%的提高。实验分析了不同切片大小对算法性能的影响,选择了1MB作为默认的切片大小。
[Abstract]:In order to improve the quality of service, reduce the waiting delay and improve the user experience, video cooperative storage is combined with network coding technology to maximize the local hit. This paper presents a local hit maximization algorithm for video slicing cooperative storage. The video is segmented by network coding technology. Upon receiving a video request, the encoded video piece is sent to the user. After receiving enough video chips, the client can decode and get the complete video. The algorithm constructs a directed graph of resource allocation, transforms the original problem into the minimum cost flow problem on the graph, and obtains the optimization scheme within polynomial time complexity by solving the minimum cost flow problem. When the total capacity of video is larger than that of server, the algorithm still performs well. Experimental results show that the performance of the proposed algorithm is superior to that of other existing algorithms, especially when the total video capacity and the total server capacity are larger, the performance is improved by 10%. The effect of different slice sizes on the performance of the algorithm is analyzed experimentally, and 1MB is chosen as the default slice size.
【作者单位】：中国科学技术大学计算机科学与技术学院;安徽省计算与通讯软件重点实验室;
【基金】：国家自然科学基金资助项目(61371118) 中央高校基本科研业务费专项资金资助项目(WK0110000029)
【分类号】：TN948.64

【共引文献】