Network Channel Coding for Packet Video

广泛的端到端信道编码技术已经用于支持因特网和其它分组网络上的分组视频应用。然而，传统的端到端前向纠错（FEC）技术不能很好地扩展，特别是对于大型网络，这是由于底层分发树的深层节点处的大量分组丢失。与此同时，新兴的网络模式，如覆盖，对等和ad-hoc无线网络，打开了大门，提供实时视频到大量的接收器，同时保持高水平的可靠性和吞吐量的新方法。在本研究项目下，一个新的框架，网络信道编码（NCC），利用新兴的网络模式，如对等和覆盖网络，进行了研究和开发。这项研究优化了NCC的通用网络应用程序，重点是优化数据包视频应用程序，其整体质量和吞吐量的大量接收器。特别地，用于在任意网络拓扑（即，任意图形）正在被研究、分析和设计。NCC优化正在使用速率失真（RD）的框架下，最大限度地提高吞吐量，同时最大限度地减少整个网络的NCC开销的成本最小化。该框架揭示了关键的理论限制和相应的最佳NCC解决方案，可以实现最大吞吐量和最小失真的分组视频应用。此外，正在开发实用的分布式NCC算法。虽然一般的性质，这些分布式NCC算法非常适合于分组视频对等和ad-hoc无线系统。此外，在优化NCC框架的背景下，正在探索和开发新兴领域中的新扩展，例如LDPC码、跨层通信和网络（源）编码。

A wide range of end-to-end channel coding techniques have been used in support of packet video applications over the Internet and other packet networks. Traditional end-to-end Forward-Error-Correction (FEC) techniques do not scale well though, and especially for large networks due to the large number of packet losses at deep nodes of the underlying distribution tree. Meanwhile, emerging network paradigms, such as overlay, peer-to-peer, and ad-hoc wireless networks, open the door for new approaches of delivering real-time video to a large number of receivers while maintaining high level of reliability and throughput.Under this research project, a new framework, Network Channel Coding (NCC), that exploits emerging networking paradigms, such as peer-to-peer and overlay networks, are investigated and developed. This research optimizes NCC for generic network applications with emphasis on optimizing packet-video applications, their overall quality, and their throughput to a large number of receivers. In particular, optimum solutions for embedding and distributing channel coding functions within arbitrary network topologies (i.e., arbitrary graphs) are being investigated, analyzed, and designed. NCC optimization is being developed using cost-minimization under a rate-distortion (RD) like framework that maximizes throughput while minimizing the NCC overhead throughout the network. This framework reveals key theoretical limits and corresponding optimum NCC solutions that can be achieved for maximum-throughput and minimum-distortion packet video applications. Furthermore, practical and distributed NCC algorithms are being developed. Although general in nature, these distributed NCC algorithms are well suited for packet video over peer-to-peer and ad-hoc wireless systems. Furthermore, novel extensions in emerging areas, such as LDPC codes, cross-layer communications, and network (source) coding, are being explored and developed in the context of the optimization NCC framework.