【作者】 张路昊；

【导师】 陈贺新；

【作者基本信息】 吉林大学 ， 通信与信息系统， 2008， 硕士

【摘要】 随着网络技术的发展与普及,网络通信的内容与形式趋于多样化。其中视频通信越来越占据重要的地位。而在各种网络视频服务类型中,实时视频传输是最有挑战性的,它对视频压缩的效率和网络友好都提出了很高的要求。在这方面H.264/AVC表现最为突出。但是视频高度压缩技术在减少冗余信息的同时,也降低了视频流抗击错误的能力。尤其是面对错误率较高而带宽受限的无线网络时,这样的矛盾就更加的突出。为了解决这一棘手问题,H.264引入了各种容错工具,如灵活宏块排序,率失真优化和冗余片结构等,但是使用这些抗误码工具提高视频传输差错控制能力的同时付出了降低了视频的压缩效率和占用了更多宝贵的网络带宽资源的代价。因此如何能根据不同的情况恰当的应用这些抗误码技术优化视频传输便成为一个重要的研究课题。鉴于国内外对于H.264的抗误码优化传输主要针对3G移动通信网和移动IP网络而较少涉及到无线局域网的情况,本文主要针对无线局域网中的H.264视频优化传输,吸收借鉴了交叉层设计思想方法,提出了一种基于无线局域网IEEE802.11MAC协议的交叉层H.264视频优化传输的方法。该方法通过统计无线局域网发送端MAC层的收包率来估计信道的状况,并根据信道估计来调整H.264编码器的参数,以达到编码参数与信道状况大致匹配的目的,优化视频传输效果。本文的实验以网络模拟的方式来实现,采用NS2仿真软件实现无线局域网的环境,将H.264视频编码文件经过解析处理后送入到模拟环境做各种情况下的传输测试,通过大量反复实验证明了该方法对信道估计的可靠性和稳定性,然后分各种信道情况和编码参数来测试视频传输质量,并以此来确定不同情况下的匹配策略,最后依照前面的实验匹配策略进行了交叉层优化传输测试,通过实验证明了它对视频传输质量的改善。更多还原

【Abstract】 With the development and popularization of network technologies, the simple information such as text, sounds and images can not satisfy user’s requirements, the transmission of video and multimedia information on network is a new kind of trend and need. Among all kinds of network video services, real-time transmission of video data is the most challenging task, as it severely requires video compression efficiency and network friendly design, but H.264/AVC just satisfy the two kinds of requirements above at some extent. H.264/AVC is the newest international video coding standard, jointly developed by groups from ISO/IEC and ITU-T, which aims at achieving improved compression performance and network-friendly video representation for different types of applications including real-time video communications. It performs significantly better than the available ISO MPEG-4 Part 2 standard and ITU-T Recommendation H.263 in terms of compression and network adaptation. However, it cuts down the ability of error resiliency as using high compression technology on video coding. The contradiction between compression efficiency and the ability of error resiliency are serious, especially facing with wireless network, which with the character of high bit error rate and limited bandwidth. In order to solve the serious problem, kinds of error control schemes are employed by H.264/AVC, such as flexible macroblock ordering (FMO), rate distortion optimization(RDO) and redundant slice(RSs). Nevertheless, it reduces the video compression efficiency and costs more precious network bandwidth resources at the same time of using these error control schemes in order to enhance the ability of error resiliency. Consequently, it becomes the focus how to adopt appropriate video coding mode and error control method. Moreover, it is the aim and target of this thesis.Nowdays, at home and abroad, Most researches on the wireless transmission of H.264 are on the context of 3G mobile telecommunication network or wireless broadband IP network, and optimize it from the respect of codec and network protocol respectively. During these years, many research institutes and researchers invite the application of cross layer design(CLD) idea to the wireless transmission of video, which solve the problem of supportless between the optimization of codec and network protocol. Here in this thesis we employ the method of cross layer design, aiming directly at the character of WLAN’s MAC layer, utilizing the frame ACK to estimate the error rate condition of the wireless channel. And then direct the latter video coding according to the result above, which achieve the target of self-adaptive adjustment of the codec according to the condition of channel. It is a cross layer design scheme that initiate from MAC layer and application layer(codec) responding to it. We use the mode of network simulation in this experiment with the simulate software of NS2 to construct WLAN for the simulation of video transmission, and then analyze the result according to the simulated experiment.In the theory part of this thesis, first of all, the coding characters of H.264 are analyzed, and the changes on network friendly design of H.264 and its significance are also introduced. Then some kinds of error control schemes and video optimization methods of H.264 are analyzed, which mainly focus on the unique error control schemes of H.264, such as flexible macroblock ordering (FMO), rate distortion optimization(RDO) , Parameter Sets and so on. Then in the next part of the thesis, WLAN is simply introduced, including the classification of topology, frame category and the structure of all kinds of frame. The thesis analyzes the characters of IEEE802.11b MAC layer’s protocol of WLAN. These are the theory foundation of the cross layer design proposed in this thesis. In the respect of network simulation, besides describing the features, basic structure and basic principle of NS2, the thesis analyzes the code implementatiom of network components involving in the experiments. There are the subject of the introduction of the video stream to the traffic generator in application layer of NS2, the subject of some kinds of timer and the subject of procedures realization of IEEE802.11b MAC layer’s protocol, especially for the last subject. These are the procedure code foundation of the cross layer design proposed in this thesis.At last, the network simulation experiment is carried out using NS2. The operating system of the experiment is windows XP+cygwin, and the edition of the codec of H.264 is JM7.2, the transmission mode is point-to point wireless transmission. The experiment utilizes the ACK frames of MAC layer to calculate the received rate of the packet at the MAC layer, and then estimates the channel condition of bit error rate according to the calculation result of the packet received rate. First of all, the experiment is about the homologous relationship between packet received rate at the MAC layer and channel condition of error bit rate. We set the coding parameter SliceArgument=100 and SliceArgument=200 respectively. The experiment carried out under several discrete bit error rate conditions and the same experiment is repeated 10 times each. From the experimental results, it is obvious that under the same channel condition of error rate, the packet received rate at the MAC layer trend to the same. Furthermore, it can distinguish each other under different error rate channel conditions. Then makes all the other coding parameters fixed except for the coding parameter RDO. The experimental results show that the packet received rate in MAC layer are not fluctuate though RDO is changed. It proves that RDO’s change will not disturb the judge of the MAC layer’s estimation of the channel conditions, which is convenient to continuous survey of the channel condition using latter coding frame that is adapted by cross layer design. The explanation of the experiment result is: if the SliceArgument is fixed, the slice length of the H.264 coding is around the value of the SliceArgument. Furthermore, the codec here we use adopt the method of simple packet scheme, which means that one RTP packet load one slice, and in other words, the RTP packet length is fixed by and large. That is the reason why the packet received rate is regular. The next step of the experiment is decoding the video which is transmitted on the WLAN and working out matching schemes. The results of two respects are mainly considered to work out the schemes: the PSNR value of the video decoded and transmission defer. At the premise of the transmission defer be up to standard, we choose the coding scheme whose PSNR value of decoded is the best. At last, we carry out the test of the cross layer design scheme of this thesis. The experimental results show that the scheme can optimize the H.264 video transmission on some channel conditions. Besides implementing optimization by adjusting the parameter of RDO, we also realize the optimization by adjusting quantization step QP. It means adjusting the coding precise extent according to the channel conditions and optimize the video transmission. In this experiment, the optimization extent is tested through adjusting QP value, which is mainly based on the condition of the channel improvement. The experiment process is similar to the RDO adjustment and better expermental results are also obtained.At last, we summarize and give some prospect of the thesis. The shortcomings of the experiment are pointed out and further improved schemes are put forward. The cross layer design scheme adopted in this thesis can be utilized together with other optimization methods of video transmission and other cross layer design schemes, but the result is not clear and need further experiments.更多还原