【作者】 于红妍；

【导师】 迟学芬；

【作者基本信息】 吉林大学 ， 通信与信息系统， 2007， 硕士

【摘要】 随着无线IP接入技术的日趋成熟、无线网络带宽的不断提高,视频流在宽带无线IP接入网中的传播已成为现实。与一般的数据业务不同,视频流的播放有严格的实时性要求,这就要求网络为视频传输提供足够的带宽、有保障的时延和误码率。与之相对,承载视频业务的无线网络却具有带宽波动、时延和时延抖动大、差错率高等弱点。同时,传统的视频业务在传输层采用UDP协议会造成对TCP流的不公平现象,并且加重网络的拥塞。因此如何让视频流在无线网络中更好的传输、最大限度的保证视频播放质量成为了业界研究的热门问题。以解决无线视频业务存在的问题为出发点,本文首先分析了MPEG-4视频流层次结构,并在此基础上提出了以视频图像重建时VOP重要性为依据的视频流细粒度等级划分算法;其次采用TFRC(TCP Friendly Rate Control)协议对视频流的发送速率进行控制,使视频流具有对TCP的友好性,并结合视频流等级划分算法实现内容敏感的视频速率控制机制,最大限度的保证了接收端的视频播放质量;随后针对传统的TFRC在无线网络中存在带宽利用率差的缺点,提出了对TFRC加入LDA(Loss Differentiation Algorithm)算法,使其能区分产生丢包事件的原因,从而提高了带宽利用率,形成适用于无线网络环境下的无线TFRC;最后,本文提出了结合视频流等级划分与无线TFRC速率控制机制的视频业务传输体系,仿真结果表明本系统不但实现了对TCP流的友好性,同时最大限度的保证了接收端的视频质量,在无线信道误码情况下亦具有很高的带宽利用率。更多还原

【Abstract】 With the development and popularization of Internet technologies, the simple information such as texts, images and sounds can not satisfy users’requirements, and these will be replaced by a kind of new multimedia information which is composed by audio, video and data. With the gradually maturity of wireless IP access technology and continually development of wireless network’s bandwidth, the propagation of video streaming through wideband wireless IP access network has come true. Different with general data service, the broadcast of video streaming call for exact real time demand, which will ask for the network afford sufficient bandwidth, ensured delay and error rate. Relatively, video services over wireless network have some shortcomings such as fluctuating bandwidth, long delay and high error rate. At the same time, traditional video services adopt UDP protocol on transport layer will be unfair to TCP streaming, also aggravate network congestion. Therefore, how to make video streaming transmit better and guarantee video broadcast quality utmost come to the fore of industrial research.In the paper, the author first analyzed the system structure of MPEG-4 video streaming, and then proposed a classification algorithm of MPEG-4 video streaming based on it. MPEG-4 video coding standard has provided video image reconstruction technologies based on objects, and coding different AVO separately is allowed. A MPEG-4 video streaming contains one or more video objects (VO), each video object sequence includes many layered video streaming, which can be expanded, each layer of video streaming is a group of VOP sequence, VOP is the basic cell of video data. What should be paid more attention through video streaming transmission is not only controlling the transfer speed of video streaming to fit the characteristics of wireless network, but also ensuring the image quality of receiving terminal. Due to MPEG-4 video streaming is composed of many objects, many expanded layers and many different types of VOP, and these different components have different influences on image reconstruction quality of receiving terminal, therefore, MPEG-4 video streaming should be classified before improving the protocol of video service, then corresponding protocol optimization should be made based on the classification result .At transport layer, traditional video operation generally adopts UDP protocol which doesn”t have congestion control mechanism. It’s easy to lead network congestion brought by video streaming. When network congestion happens, UDP protocol doesn’t realize the current status of the network congestion. It goes on sending large numbers of data. So it’s difficult to resume from the status of the network congestion.At the same time, because of the AIMD speed control tactic (addition increase, multiplication decrease) of the TCP protocol, TCP streaming in the network will decrease the sending quantity. In this way, it will abate the status of the network congestion. But due to the presence of the UDP streaming, the status of the network congestion can’t be improved. TCP streaming has to go on releasing the bandwidth. It finally leads that the important operation which adopts TCP protocol can’t be transmitted. The unfair phenomenon to TCP will occur. Therefore, it’s vital to add TCP-Friendly congestion control mechanism in video transmission.This paper adopts TFRC(TCP-Friendly Rate Control) to control the speed of the video streaming. TFRC is current more mature TCP-Friendly control protocol controlled by the sending end. It became international standard RFC3448 of the RFC in January,2003. In wire environment, TFRC can recognize the status of the network congestion by losing package event which is discovered at sink. Thereby, the sending speed can be adjusted based on the network status. And finally the aim to TCP-Friendly[0] is achieved. But in wireless environment,due to the high channel error rate,the losing package event happened at sink not only because of the network congestion,but also the channel error code . But TFRC protocol mechanism which was setted in RFC3448 cann’t distinguish the reason of these two losing package. When the losing package happens, it is usually thought that It is caused by the network congestion. Thereby, the sending speed is reduced mistakenly .It leads the fall of the bandwidth using rate. Aim at this problem, the paper lodge adding LDA arithmetic at TFRC sink[0]. It makes TFRC to distinguish the procreant reason of the losing package event. It improves the bandwidth using rate.Combined with the classification of video streaming and wireless TFRC rate control scheme, this paper proposes a system framework of video transmission. In the sending terminal, MPEG-4 video streaming is sent to classification module is to do VOP classification. And according to the current sending rate returned from wireless TFRC protocol module, the VOP packets of[0] high priority are chosen to send. Wireless TFRC protocol module calculates the current sending rate based on the feedback information sent from receiving terminal, submits it to the classification module. At the same time, the VOP packets which are selectivity sent by the classification modual will be timestamped, and then sends it to the network for transmission by UDP protocol. Wireless TFRC module of receiving terminal first processes TFRC packets, picks up the timestamp and the packet number from them, differentiates the reason of packet loss events and calculates loss event rate. Then it sends feedback information to the sending terminal regularly and put the video data into the buffer for synchronization simultaneously. Finally the video image is reconstructed by decoder.Because of the high cost of constructing a real network to experiment and many nonideal scene factors in real network, the real effects from all kinds of factors are understood hardly in experiments. So this paper adopts NS network simulation software to do simulation experiments. In the aspect of video traffic simulation, this paper uses the method of drawing the Trace file of video streaming into NS simulation environment, which can reflect the change of video traffic exactly and be convenient to mark each VOP priority level simultaneously. The network topology used in experiments is wireless LAN topology which is the combination of wire and wireless. Because wired and wireless links coexist, hierarchical routing of network should be set up and addresses should be allocated for each node.It has four emulate research in this paper. First one has validated that the feasibility of real video streaming Trace file emulation method, and has analyzed the reason that video quality at receiving terminal was poor without classification of video streaming algorithm. Second one has validated that video quality at receiving terminal was worse than wired network in wireless network topology structure without channel error. Third one has contrasted that UDP was inequitable to TCP with that TFRC protocol was friendly to TCP, and the infection to video quality at receiving terminal with or without classification of video streaming. Fourth one has validated that TFRC has an disadvantag that was the decrease of bandwidth utilization ratio with wireless transimission losses, but improved wireless TFRC enhanced the bandwidth utilization ratio clearly with wireless channel error.The work has been done in the pape[0]r is summarized at the end, and put forward further improved scheme. Classification of video streaming algorithm in this paper can be applied to the optimization of protocol in different layers[0]. It has only discussed that the application of transport layer protocol optimization in this paper, and the wireless video transmission system more perfect combined with services division model in network layer and unequaled error protection mechanism joint source channel coding.更多还原