Efficient Integration and QoS Management of Video over Wireless Networks

我们目前正在见证我们获取信息和计算能力的方式发生了令人难以置信的变化。随着个人数字助理、蜂窝电话和掌上电脑等小型手持设备的出现，用户经常随身携带他们的计算机。此外，随着无线技术的发展，用户开始慢慢摆脱对有线访问信息的依赖。虽然这一趋势正在形成显著的势头，但通过这些设备访问的信息类型仍然严重受到无线媒体的限制。考虑到当今功率受限的便携式设备的瓶颈，用户还没有感受到信道条件的限制；然而，底层无线网络的压力只会变得更糟。提出的研究目标是构建高性能的无绳接入算法，通过跨多层的联合优化来考虑系统的整体目标。目标将是解决将物理层连接到应用层的问题。在过去，无线通信系统设计采用了一种经典的方法{一种基于开放系统互连参考堆栈模型的模块方法。对于有线通信网络，这样的活动分离导致了提供可靠性能的正常运行的系统。然而，无线通信信道的有害影响迫使人们研究所有可能的方法来提高性能。因此，为了提供所需的保真度并在无线介质上传输高数据速率信息，跨多个通信层的联合优化将是必要的。为了集中研究视频的无线传输，将考虑直接序列码分多址(DS-CDMA)的无线传输。在这个项目中，我们提出了使用DS-CDMA作为接入方法来研究和开发更适合无线传输的视频流算法。有几个理由说明这个项目是必要的。首先，目前物理层的多速率检测和估计算法的研究并不是针对特定的应用层数据定制的。其次，许多无线网络上的视频传输方案都假设了无线传输引入的固定误码率。在码分多址系统中，误码率高度依赖于信道和活跃用户的数量，这两个过程都是时变的；然而，在应用层还没有完全解决误码率的这种变化。第三，可以利用视频的弹性来设计高效的算法，该算法可以利用视频流的形状以及时变的信道条件来提供所需的服务质量。本研究项目有三个主要目标：研究无线物理媒体传输的高效多速率检测和估计算法；研究无线网络中的视频流技术，该技术利用底层网络状况反馈来动态调整用于保护视频数据的前向纠错；研究物理层媒体特性和更高级别的视频应用的集成，以向应用提供最高质量的服务。

We are currently witnessing an incredible transformation in the way we access information andcomputing power. With small handheld devices such as personal digital assistants, cellular phones,and palmtop computers, users are taking their computers with them more often than not. Fur-thermore, with the developments in wireless technologies, users are beginning to slowly move awayfrom their dependence on wired access to their information. While this trend is building significantmomentum, the types of information that are accessed through these devices is still severely lim-ited by the wireless media. Given the bottle-necks associated with today's power-limited portabledevices, the limitation due to channel conditions has not yet been felt by the user; however thestrain on the underlying wireless network will only get worse. The goal of the proposed research is the construction of high performance algorithms for tetherlessaccess which consider overall system goals through joint optimization across multiple layers. Theobjective will be to solve problems which connect the physical layer to the application layer. In thepast, a classical approach to wireless communication system design was undertaken { a modularapproach loosely based on the Open Systems Interconnection reference stack model. For wiredcommunication networks, such a separation of activities led to functioning systems offering solidperformance. However, the deleterious effects of the wireless communications channel force theinvestigation of all possible methods to improve performance. Thus, to provide the desired fidelityand to transport high-data rate information over the wireless medium, joint optimization acrossmultiple communication layers will be necessary. To focus the research, wireless transport of videovia Direct Sequence Code-Division Multiple-Access (DS-CDMA) will be considered. In this project, we propose the research and development of video streaming algorithms that aremore suited for wireless transmission using DS-CDMA as an access methodology. There are severalreasons why this project is necessary. First, current research onmulti-rate detection and estimationalgorithms at the physical layer are not tailored for specific application layer data. Second, many ofthe video transmission schemes over wireless networks have assumed fixed bit-error rates that areintroduced by the wireless transmission. In CDMA systems, the bit-error rate is highly dependenton the channel and the number of active users, both of these processes are time-varying; however,this variability in bit-error rates has not been fully addressed at the application layer. Third, theelasticity ofvideo can be exploited to design efficient algorithms which can take advantage of theshape of video streams as well as time-varying channel conditions to offer desired quality-of-service. This research project has three principle goals:Investigation of efficient multi-rate detection and estimation algorithms for wireless physicalmedia transmission.Investigation of video streaming techniques for wireless networks that use underlying networkcondition feedback to dynamically adjust the forward error correction used to protect thevideo data.Investigation of the integration of physical layer media characteristics and higher-level videoapplications to provide the highest quality-of- service to applications.