【作者】 王永芳；

【导师】 余松煜；

【作者基本信息】 上海交通大学 ， 通信与信息系统， 2007， 博士

【摘要】 无线视频传输的QoS技术涉及到视频编码、无线通信、优化理论、网络结构和网络管理等许多研究领域的核心技术。针对目前视频在实际网络传输中所面临的难以提供QoS保证的问题,本文主要研究了视频在无线网络传输中的鲁棒性以及所用资源的优化分配。在充分利用相关技术优点的基础上,对无线视频的QoS在鲁棒性、稳定性和适用性上的不足之处进行了深入研究,提出了新的处理方法和模型,从而有效地提高了视频传输的QoS。首先,在探讨功率消耗和视频传输的QoS之间的权衡关系基础上,本文提出了行之有效的解码端功耗优化的视频传输模型。针对Turbo码的解码特点和视频解码优先级的数据分类方法,设计了一种视频传输系统参数优化配置方案-不均等解码功耗分配优化算法。该算法实现了视频QoS和功率消耗的权衡,能够在功耗有限的情况下,获得更好的视频QoS。其次,为了高效利用有限的网络资源,实时视频流量预测是必需的,本文提出了一种新的基于H_∞算法的视频流量预测模型。该模型不仅克服了传统模型中固有的缺陷,而且在预测过程中无须知道先前的噪声统计特性,同时又借鉴了H_∞算法本身鲁棒性的性能。通过对视频流量预测结果的定量分析,表明该模型预测准确,稳定性高,且具有很好的鲁棒性。最后,提出了两种新的提高视频QoS的鲁棒性传输系统,即不均等分集增益的LSTBC_OFDM_LDPC视频传输系统和基于LDPC的不均等错误保护的视频传输系统。LSTBC_OFDM_LDPC视频传输系统不仅通过LSTBC的分集增益和编码增益以及LDPC的编码增益提高了系统的性能,而且本文利用系统提供的不均等分集增益去匹配不同优先级的视频流,提高了视频传输的效率。此外,利用LDPC编解码的优越性,本文提出了基于LDPC和数据分类的不均等错误保护的视频传输系统。通过联合信源信道编码的优化比特分配,获得了最好的视频重建质量。实验结果证实,在相同的带宽下,提出的基于LDPC的不均等错误保护方法与均等错误保护方法相比,PSNR提高了大约0.5-3dB。更多还原

【Abstract】 Wireless video QoS technique is a multidisciplinary topic involving many areas, ranging from video coding, wireless communication, optimization theory to networking architectures, networking management and so on. Considering the guaranteed QoS is difficult to achieve for wireless video delivery, the robustness and optimization resource allocation are investigated in this dissertation to provide better video QoS. By taking full advantage of the technologies in the related application, we have done further study on their shortage of robustness, stability or applicability and proposed some novel processing methods and models to efficiently improve video delivery QoS.Firstly, based on the tradeoff between the power consumption and end to end distortion, we propose an efficient decoding power consumption optimization framework. By exploiting turbo decoding and source priority, an unequal decoding power allocation algorithm is proposed, which results in an excellent tradeoff between video quality and power consumption and yields significant improvement of video quality under constrained power consumption.Secondly, for the efficient utilization of network resources, video traffic prediction is necessary in the real application. A novel model based on H_∞algorithm is proposed for video traffic prediction, which not only overcomes the inherent defects in traditional methods but also does not require a priori knowledge of the noise source statistics. More importantly, the proposed prediction model makes use of H_∞model to improve its robust performance. The quantitative analysis of the video traffic prediction results indicates the model has more accurate prediction, better stability, and higher robustness.Finally, two robust video delivery systems are proposed to improve wireless video QoS, namely, unequal spatial diversity gain transmission system based on LSTBC_OFDM_LDPC and unequal error protection video transmission system based on LDPC. The first proposed system not only significantly improves the system performance by exploiting both the spatial diversity gain of LSTBC and the coding gain of LDPC, but also provides an unequal diversity gain mode to match the source of different prioritized streams to achieve better video transmission performance; The second proposed system, by jointly considering source and channel bit allocation under bandwidth constraints, features an unequal number of channel rate bits assignment to different priority stream. Simulation results demonstrate that the proposed algorithm is more robust compared with those obtained with equal error protect (EEP) technique and can achieve 0.5-3 PSNR gains.更多还原