【作者】 汪淑娟；

【导师】 梁满贵；

【作者基本信息】 北京交通大学 ， 信号与信息处理， 2013， 博士

【摘要】 随着无线宽带技术的快速发展,无线网络的容量有了很大提升。各种差错控制技术的不断改进也为无线网络的可靠传输提供了一定保障。这些技术的结合使得无线通信网络的业务承载能力大大增强,为诸如流媒体传送等新兴服务类型的出现提供了契机。尤其是可扩展视频编码技术的提出和标准化,在解决无线多播的异质终端问题的道路上迈出了一大步,它使得源端可以通过一次多播满足多个不同终端的数据接收需求,能够真正发挥多播的优势。然而,无线网络上的宽带多媒体业务的推广仍然存在着许多挑战：一方面,尽管网络的容量有所增加,其增加速度仍然无法与多媒体业务对带宽的需求增长速度相比,相对于旺盛的市场需求,带宽仍然是制约无线多媒体业务的一个重要因素；另一方面,尽管可以在传输中采取诸如自动请求重传、不等错误保护以及前向错误纠正等差错控制手段保护传送的数据,由于无线信道先天的物理特性,丢包仍然是不可避免的。重传所丢失的数据包所带来的时延会对用户的服务体验造成重大影响。与设计普通业务在无线网络中的传输机制不同的是,包时延、抖动、或者丢包率等传统服务质量(Quality of Service, QoS)参数并不能准确的刻画多媒体业务的实际传输质量,用户的主观体验质量(Quality of Experience, QoE)才是评价传输效果的重要依据。而用户的服务体验质量是多个复杂因素共同作用的结果,很难映射为具体的技术参数。因此,在设计无线多媒体业务传输机制时,需要同时考虑具体应用的传输需求和用户对传输效果的要求,利用有限的无线网络资源,为用户提供优质体验的多媒体数据传送服务。研究证实,网络编码技术能够增大多播网络容量,缩短重传时延。因此,在多媒体业务传输过程中恰当的利用网络编码技术能够提高传输效率。本文针对无线网络多媒体数据传输的具体应用场景,讨论了利用网络编码技术实现数据的高效可靠传输,为终端用户提供满意的服务体验的问题。论文首先研究了利用网络编码优化按需数据分发系统的接入时间的问题,提出一个启发式算法对所有用户的接入时间进行优化；然后,研究了可扩展视频业务的传输优化问题,对可扩展视频流的重传时延的优化是一个混合优化问题,我们也提出一个启发式算法进行求解；最后,受可扩展视频的优化问题的启发,我们对一类具有解码依赖关系的有序数据的传输优化问题进行了研究。主要的研究内容和创新点包括如下几个方面：1)研究了按需数据分发业务的用户接入时间最小化问题。通过分析多个用户所请求的数据之间存在的相关性,提出一个启发式的编码算法。该算法在每次多播前贪婪的搜索能满足最多用户的数据集合,并对该集合中的包进行线性编码后发送给用户。与直接采用随机线性编码算法进行多播相比,所提出的算法充分考虑了用户的实时收包情况,在普遍情况下具有更好的时延性能；只有当所有用户请求的数据和所接收到的数据的分布情况一致时,直接采用随机线性编码算法才具有较好的时延性能。2)研究了无线网络多播可扩展视频的视频重建时间最小化问题。从可扩展视频流的结构特征出发,提出了一个两阶段的可靠传输方案。(a)第一阶段是多播阶段,由于可扩展视频的所有增强层的解码都依赖于基层的正确解码,引入了层次化网络编码技术对视频的基层提供保护,保证在信道条件不佳的情况下,用户仍然可以有一定的概率解码基层,获得基本的重建视频质量。(b)第二阶段是重传阶段,根据第一阶段用户的收包情况和用户对视频重建的质量要求,重传视频包。提出一个最大化可解码用户数的编码算法,优化用户恢复视频所需要的等待时延。所提出的算法在重传过程中能够为用户提供一种“软”的视频恢复效果,视频重建的质量随着收包数的增加而提升。此外,算法还兼顾了各视频用户之间的公平性,具有低恢复质量要求的用户可以比高要求的用户更快的得到满足。3)研究了有序数据的渐进式重传问题。有序数据指的是一类具有优先级的数据,低优先级数据的解码依赖于高优先级的数据的解码。因此,传输此类数据时需要考虑数据在接收端的实际解码时延。(a)研究了理想反馈条件下有序数据的重传问题。采用包权重刻画数据间的优先级以及解码依赖关系,在此基础上,提出了两个优先级编码算法。所提算法能够为用户提供即时解码功能,使用户可以边接收包边解码,渐进的恢复原始数据。此外,用户也可以根据自己的需要制定接收结束时刻。与文献相比,所提出的算法对应特定优先级数据的解码时延更短,且系统的总接入时延也更小。(b)研究了非理想反馈条件下有序数据的重传问题。在已有工作的基础上,针对用户反馈丢失的情况,设计了相应的包更新策略,提出了非理想反馈的包选择编码算法。大量的实验结果表明所提出的算法能够较好的抵抗用户反馈丢失的事件,与文献相比,同等条件下具有更短的解码时延和总接入时间。无线网络多媒体通信方兴未艾,用户的服务体验是能否吸引订单的一个决定性因素。本论文从技术和用户需求角度出发,考虑在多种特定应用场景中优化用户的等待时延,提升用户服务体验。所提出的算法一方面能够更快的使用户接收到所要求的数据：另一方面还为用户提供了一个功能,使用户可以根据自己情况,“定制”服务,当用户主观上已经满意时,可以主动中断数据传输。这一功能贴合了无线用户的异质性要求,在信道带宽有限时也可以为系统节省一定资源。更多还原

【Abstract】 With the rapid development of wireless broadband technology, the vol-ume of wireless network has been greatly improved. Meanwhile, the error control meth-ods provide good support for robust wireless communications. The combining of large capacity and efficient error control methods enhances the wireless network’s ability to provide appealing applications, for example, the wireless streaming services. Moreover, the scalable video codec has been proposed and standardized to meet the requirements of heterogeneous wireless users. The scalable video coding standard enables the source to satisfy all users through one multicast and thus makes the best use of wireless multicast. However, the exploiting and spread of wireless broadband applications still face many challenges:On one hand, although the network capacity is growing, it is not fast enough to meet the expanding need of market. The bandwidth limit still hold the whole market back. On the other hand, owing to the physic nature of wireless channels, the packets erasure is inevitable no matter what error control method we adopt. The retransmission delay can be a great factor for the quality of experience (QoE). Differ from the design of wireless transmission strategies for common applications, the old metrics of Quality of Service (QoS), such as packet delay, jitter and packet loss ratio, are no longer accurate to evaluate the transmission of multimedia data, and the QoE becomes a much powerful tool to score the multimedia transmission. However, QoE is a result of multiple factors that is difficult to map to specific parameters. When designing the wireless transmission strategy for multimedia applications, the transmission requirements posed by particular application as well as clients should be considered at the same time. A good strategy shall is able to provide transmission service with high QoE with limited resources. Net-work coding has been proved to be an efficient way to improve the capacity of wireless multicast and to reduce the retransmission delay. Thus, it is beneficial to apply network coding appropriately in the transmission process for multimedia applications. This thesis considers the problems that applying network coding to transmit multimedia data effi-ciently and to provide good QoE for end users in wireless network. First, we studied the optimization of access time in on-demand data dissemination system based on network coding, and proposed a heuristic algorithm to minimize the total access time. Second, we studied the optimization of scalable video transmission. The problem is a combinatorial optimization problem and we designed a heuristic algorithm to resolve it. Last but not least, inspired by the second optimization problem, we studied the optimization problem of progressive transmission for in-order data.The main research innovations are as follows:1) Studied the problem of minimizing the total access time in on-demand data dis-semination system. Analyzed the relevance between multiple users’ requested data items and proposed a heuristic algorithm. The proposed algorithm searches the best data items set that can meet the most users’ requirements, and conducts linear com-bination on the found data items before multicast. Comparing to the random linear network coding method, our algorithm considers the users’ instant reception status when searching the coding set, and thus achieves better delay performance in most cases, only when the reception status of all users are the same and they request the same data items that the random linear network coding method works better.2) Studied the minimization of video reconstruction time for scalable video in wireless network. Based on the structure characteristics of scalable video, proposed a two-stage transmission scheme.(a) The first stage is the multicast stage. All the enhancement layers of scalable video cannot be decoded until the base layer is decoded successfully, hence, the original video packets are encoded by hierarchical network coding method in order to provide some protection for the base layer, and to guarantee the base layer decodable probability.(b) The second stage is the retransmission stage, the sender retransmit packets based on receivers’ requirements for video reconstructed quality and the pack-ets reception status in the first stage. Proposed a maximizing decoding users algorithm to optimize the waiting time receivers experience to get enough packets to reconstruct the video. The proposed algorithm provides receivers a "soft" video recover process that the quality of reconstructed video improves with the receiving of video packets. Further, the proposed algorithm exhibits a "fairness" among receivers that low-demanding receivers wait shorter than high-demanding receivers.3) Studied the progressive retransmission of in-order data. The in-order data represent that data with priority that the decoding of low priority data depends on the decod-ing of high priority data. Thus, it is important to consider the practical decoding delay on the receiver side for the transmission of in-order data.(a) Studied the progressive retransmission of in-order data under assumption of perfect feedback. The packet’s priority and the decoding dependency of in-order data are addressed via weighting in the coded packets in the construction process to maximize a utility function. Proposed algorithms enable receivers to decode the retransmission packets instantly and progressively recover the original data, such that receivers can terminate the retransmission process at any point. Compared with the algorithm proposed in reference, our algorithms have shorter completion time and total access time.(b) Studied the progressive retransmission of in-order data under assumption of imperfect feedback. Based on the accomplished work, designed appropri-ate packets updating schemes and proposed corresponding coding algorithm-s. Extensive simulation results justify that the proposed algorithms achieve shorter completion time and total access time and thus fit for the imperfect feedback scenarios.The multimedia communications over wireless networks is rising, and the quality of experience of users is the key to its success. This thesis aims at optimizing delay and providing good QoE in multiple practical scenarios. The proposed algorithms not only serve users with better data transmission characteristics, but also provide an option that users can "customize" the service and cut off the transmission as long as the received data are enough for supporting their expecting QoE. These properties of proposed algorithms fit in with the heterogeneous property of wireless networks, and network resources can be saved when bandwidth is competed.更多还原