【作者】 赵增华；

【导师】 舒炎泰；

【作者基本信息】 天津大学 ， 计算机应用技术， 2006， 博士

【摘要】 随着无线通信技术的发展,视频通信的需求日益强烈。然而,无线网络固有的传输特性,给视频传输带来巨大挑战。可伸缩视频编码是目前最有发展前景的、符合无线网络传输特性的编码方案。本文将可伸缩视频编码和无线网络传输技术有机结合,采用跨层设计理念,较全面地研究了无线网络视频传输面临的关键问题。首先,在H.264/AVC可伸缩扩展的框架下,提出了MCTF的性能改进方法,包括增加更新操作的约束条件和残差图像的去块滤波操作。在JSVM0上的测试结果表明,当视频中的运动比较剧烈并且出现很多块效应时,改进的MCTF能取得很好效果,而且编码效率也有一定程度的提高。其次,提出了一种分形和小波相结合的混合视频编码算法FEZW。该算法在EZW编码的基础上,采用了小波域的分形预测方法。实验结果表明,FEZW能提供可截断性码流;与EZW相比,在低码率下,FEZW恢复视频的PSNR值有所提高;在丢失相同的数据时,FEZW的PSNR值下降速度较EZW慢,上述特点非常适合于无线网络中的视频传输。再次,根据无线网络视频传输的率失真模型,提出了跨层的自适应速率控制机制CLARC。该机制和可伸缩视频编码相结合,利用应用层和网络层的反馈信息调整视频发送速率。仿真和实验结果均表明,CLARC算法能有效区分拥塞丢包和网络差错丢包;能适应网络带宽的波动;对网络拓扑变化反应敏感;能和TCP业务共存。另外,该算法速率控制较平稳,没有剧烈波动,在速率和总体失真二者间取得了较好的折衷,恢复视频的主、客观质量较好。然后,应用网络流理论,给出了多数据流拥塞优化路由问题的数学描述,在DSR的基础上提出了启发式算法CO-DSR。仿真结果表明,CO-DSR可以提高网络的端端吞吐率。该算法和应用层的速率控制CLARC相结合,能提高FEZW视频流的综合传输性能。最后,基于WRR,提出了一种视频业务调度算法WRR-CSDPS。该算法和DCF协议紧密结合,能充分利用信道状态调度数据包。仿真结果表明,WRR-CSDPS提高了实时业务的端端吞吐率,降低了端端延迟,延迟抖动和丢包率,可以在一定程度上保证实时业务的服务质量。更多还原

【Abstract】 Video application is becoming an important application under the next-generation wireless networks. However, supporting video communication over wireless networks poses many challenges due to fluctuations of wireless channel conditions. Scalable video coding has been shown to be capable of coping effectively with time-varying wireless environment. In this dissertation, we proposed a cross-layer framework to solve scalable video streaming over wireless networks.The scalable extension of H.264/AVC adopts a hierarchical prediction structure based on open-loop MCTF (Motion-Compensated Temporal Filtering). In this dissertation, we proposed new algorithms to improve the performance of MCTF under JSVM0 framework, including update constraints, residual image deblocking filtering. We also designed a new fractal wavelet hybrid video codec based on EZW, called FEZW. FEZW inherits the embedded structure from EZW. Therefore, its bit rate can be truncated to meet the network bandwidth requirements. On the other hand, the fractal prediction method increases its PSNR at low bit rate. This is also verified via simulations.To improve the quality of received video, we proposed a cross-layer adaptive rate control mechanism, CLARC, which makes good use of 1) routing information from the lower network layer, and 2) the frame loss feedback from the application layer. The proposed scheme tries to adapt the video encoder’s output bitrate to closely mimic the available network bandwidth. CLARC is also congestion-aware, robust to random error, and can co-exist with other TCP flows in a friendly manner. Extensive simulations were performed, as well as actual experiments (on our own ad hoc video streaming testbed) were conducted to validate the effectiveness of our scheme.Furthermore, we instigated the congestion optimized routing in wireless ad hoc network and presented a CO-DSR (Congestion Optimized DSR) algorithm based on DSR protocol. The CO-DSR scheme considers both link bandwidth and the number of flows on the path when selecting a route. Simulation results show that CO-DSR can effectively improve the network throughput, the quality of the received FEZW video stream.Lastly, based on the classic WRR (Weighted Round Robin) concepts, we proposed an efficient real-time traffic-scheduling algorithm for WLAN. Our algorithm更多还原