【作者】 张军；

【导师】 李泽平；

【作者基本信息】 贵州大学 ， 计算机应用技术， 2017， 硕士

【摘要】 互联网技术的飞速发展和网络应用的不断丰富,让流媒体技术得到了广泛关注和普遍应用。与传统媒体传输和播放方式不同,流媒体用户通过终端可以快速便捷地获取想要的内容。对流媒体数据的采集、加工、存储、调度、传输等关键技术的研究,已成为音视频领域的热点。移动互联网作为传统互联网的延伸得到蓬勃发展,基于HTTP的流媒体技术也被广泛部署和应用,苹果公司的HLS技术是其中具有代表性一种方案。研究和实现基于HLS高效流媒体分发技术是本文的主要工作。针对流媒体分发过程中编码标准的多样性、通信链路状态的不稳定性、终端异构等问题,分析视频编码原理、标准以及转码技术,提出了基于开源库FFmpeg的视频多码率转码方案,该方案能为用户提供差异化、个性化、自适应的转码服务。在大型流媒体分发系统中,通常采用集群服务模式,由于需要对发布的视频文件进行多码率转码以及多副本分布式存储导致大量冗余副本文件产生。如何对这些副本文件进行有效的管理成为提高分发系统服务能力和服务质量亟待解决的关键问题。本文分别就副本对象选取、副本数量控制、副本放置三个方面进行了研究。算法1是副本数量控制算法。算法通过对副本放置和删除代价的合理评价,对副本数量进行动态增删,以提高系统资源利用率。算法2是副本对象选取算法。该算法通过计算文件支持度来表示文件热门程度,然后利用副本预取算法RPA对将来可能成为热门的流媒体文件进行副本复制,通过对这些副本文件的分布式存储来降低用户访问延迟。算法3是副本放置算法。该算法针对集群服务节点的异构问题,使用一种多指标的负载评价方法来计算服务节点的真实负载,并根据计算负载实现集群环境下副本的最优放置。通过仿真实验验证了上述算法的有效性。最后,本文搭建了基于HLS的流媒体分发系统,通过该系统验证了多码率转码方案的可行性和有效性。搭建的系统具有视频上传、切片、转码以及用户管理、视频分发等视频点播网站的基本服务功能。更多还原

【Abstract】 With the rapid development of Internet technology and the continuous enrichment of network applications,streaming media technology has been widely concerned and used.Unlike traditional media transmission and playback,streaming media users can quickly and easily access the desired content through the terminal.The research on the key technologies such as the collection,processing,storage,scheduling and transmission of streaming media data has become a hot spot in the field of audio and video.Mobile Internet as a traditional extension of the Internet to flourish,based on HTTP streaming technology is also widely deployed and applied,Apple’s HLS technology is one of the representative projects.Research and implementation based on HLS efficient streaming media distribution technology is the main work of this paper.Aiming at the diversity of coding standard,the instability of communication link state and the terminal heterogeneity in streaming media distribution,the video coding principle,standard and transcoding technology are analyzed,and the video multi-bit rate conversion based on open source library FFmpeg is proposed.Code scheme,the program can provide users with differentiated,personalized,adaptive transcoding services.In a large streaming media distribution system,the cluster service mode is usually used,due to the need for the release of video files for multi-rate transcoding and multiple copies of distributed storage led to a large number of redundant copy files.How to manage these copy files effectively becomes the key problem to improve the service capability and service quality of distribution system.In this paper,the author chooses the copy object,the number of copies control,the copy of the three aspects of the study.Algorithm 1 is a copy number control algorithm.The algorithm makes a dynamic addition and subtraction of the number of replicas to improve the utilization rate of the system resources by reasonable evaluation of the copy and deletion costs.Algorithm 2 is a copy object selection algorithm.The algorithm calculates the file support degree to represent the popularity of the file,and then uses the replica prefetch algorithm RPA to replicate the popular streaming media files in the future and reduce the user access delay by distributing the replicas of these replica files.Algorithm 3 is a copy placement algorithm.The algorithm uses a multi-index load evaluation method to calculate the real load of the service node for the heterogeneous problem of the cluster service node,and realizes the optimal placement of the copy in the cluster environment according to the calculated load.The effectiveness of the above algorithm is verified by simulation experiments.Finally,this paper builds a HLS-based streaming media distribution system,which verifies the feasibility and effectiveness of the multi-rate transcoding scheme.The system has a basic service function such as video upload,slice,transcoding and user management,video distribution and other video-on-demand website.更多还原