【作者】 易鸿宾；

【导师】 赵衍运；

【作者基本信息】 北京邮电大学 ， 通信与信息系统， 2008， 硕士

【摘要】 近几年,随着网络应用的普及,人类通过网络获得的信息越来越丰富,同时对网络的依赖程度也逐步增加。多媒体信息具有直观和信息量大等特点,受到人们的广泛欢迎,这使得网络上的各项视频业务得到了飞速发展,如网络视频点播、网络可视电话、IPTV等。在网络传输中,实际信道难以保证稳定可靠的传输质量,网络拥塞和网络误码都将导致信息的丢失。而原始视频信息数据量庞大,为了能在带宽受限的信道上传输视频,对视频信息进行数据压缩是必不可少的。为了获得高压缩比,目前的视频压缩编码技术均采用了运动补偿、预测和变长编码等技术来去掉视频序列中的大量时空相关性。冗余数据的减少使得压缩视频流在传输过程中对信道误码和包丢失十分敏感,即使单个突发性错误,也可能严重影响接收端的正常解码,造成恢复视频质量的急剧下降。因此视频业务的QoS受到了广泛关注,成为视频通信领域的研究热点之一。本文主要研究视频通信系统开发及QoS技术中的相关问题,主要完成了如下一些工作:①研究并设计实现了一个具备并发服务能力的视频点播服务器,能够提供多用户视频点播服务。对于服务器端与客户端之间的命令交互,我们采用TCP协议以保持其完整和可靠递送,对于视频数据的传输由于实时性的要求则采用UDP协议,上层先用RTP协议进行封装。②研究和实验分析了H.264/AVC中相关差错控制技术,如灵活宏块次序FMO等。利用H.264/AVC的数据分割机制,实现了一种基于RTP的视频数据分割打包算法和FEC编译码算法,并通过不均衡错误保护来提高视频通信的抗差错能力。实验表明,这种方案在保证实时性的前提下能够有效的降低丢包率,提高了视频传输的可靠性。③本文还研究并实现了一种基于接收端反馈的速率控制方法,以解决拥塞控制问题。服务器可以根据客户端反馈的丢包率参数进行决策,对发送速率进行实时调整,从而降低丢包率,缓解拥塞,提高信道利用率。更多还原

【Abstract】 As the prevalence of network applications in recent years, people has got more and more abundant information through the network, meantime they depend on network more and more. Multimedia information is welcomed by people because it is intuitionistic, full of great information content and so on. As a result, different kinds of video services on network are rapidly developed such as network video on demand, network video phone, IPTV and so on.Steady and reliable transmission quality cannot be achieved by real channel during transmission, network congestion and network bit error both could cause information loss. At the same time, raw video data has got a large amount of information, so it is absolutely necessary to compress video data so as to carry out video transmission over bandwidth limited channel. For high compression ratio, video compression encoding standards today all adopt technologies such as motion compensation , prediction, variable length codes and so on to reduce abundant space-time relativity existing in video sequence. Reduction of redundant data make compressed video stream more sensitive to bit error and packet loss, even if single burst error has happened, it will take serious affect to natural decoding of the client and the video deconstruction quality will be rapidly dropped. As a result, QoS of video services is now attracting great attention, and it becomes one of the hot topics in video communication field.This article mainly focuses on research and development of video communication system and related problems of QoS technologies, we mainly complete some tasks as follows: Firstly, we research and develop a video server with the ability of supporting concurrent streams, It can supply the multi-users with video on demand service. We adopt TCP protocol to assure the integrative and reliable delivery of alternant command, and UDP protocol is used on video data transmission as a result of its real-time requirement, meantime video data is encapsulated on the high layer using RTP protocol before. Secondly, we research and carry out experiments to analyze the related error control technologies based on H.264/AVC such as Flexible Macroblock Order and so on. Besides a scheme adopting H.264 data partition, fragmented packet encapsulation using rtp and unequal error protection based on forward error correction are presented. The experiment results demonstrate that this scheme could reduce packet loss rate effectively, also the real-time performance is assured and the reliability is enhanced. Thirdly, we research and realize a rate control method based on feedback from receiver so as to solve the congestion problem. The server could make a decision and adjust the sending speed of video packets timely according to the packet loss rate parameter submitted by client so as to reduce the packet loss rate, help to solve congestion and improve channel efficiency.更多还原