Collaborative Research: Systematic Optimization in Wireless Multicasting

摘要组播是一种将共同信息从一个源传输到多个目的地的技术，是多媒体分发、信息更新、群组会议等网络应用的核心组成部分。与传统的复制转发方法相比，在中间终端对网络跟踪进行创造性编码可以显著提高组播的吞吐量。由于无线介质的开放性，无线链路的通信吞吐量取决于其传输功率和附近网络终端产生的干扰。本研究的目标是在网路编码的框架内，透过共同优化传输功率、速率和进度，来建立一个系统的架构，以最大限度地发挥一般多播的效用。研究人员通过拓扑图对无线自组织网络进行建模，该网络包含点对点链路和点对多点超弧链路，具有耦合链路吞吐量。在网络编码最优的假设下，研究首先建立了迭代梯度导向的优化框架。将网络效用最大化问题转化为在梯度方向上最大化近似效用的传输调度问题，并结合不断更新近似效用及其瞬时梯度方向的转向矢量更新。然后，研究将该框架扩展到具有多个组播会话的网络的效用最大化。最后，研究开发了分布式算法来优化使用本地控制器的大规模网络的全局效用。除了计划中的研究之外，研究人员还将尝试将算法扩展到具有时变信道的自组织网络，其中效用最大化需要有效地利用信道分集增益。

Research AbstractMulticasting, where common information is transmitted from a source to multiple destinations, is the core component of many network applications such as multimedia distribution, information update, group conferencing, etc. Creative encoding of network trac at the intermediate terminals can significantly improve the throughput of a multicast over conventional replicate-and-forward approaches. Due to the open nature of the wireless medium, communication throughput of a wireless link depends on its transmission power and on the interference generated by nearby network terminals. The goal of this research is to develop a systematic framework for maximizing a general multicast utility function via the joint optimization of transmission power, rate, and schedule, within the framework of network coding.The investigators model a wireless ad hoc network by means of a topology graph, which contains point-to-point links and point-to-multipoint hyperarc links with coupled link throughput capacities. Under the assumption of optimal network coding, the research first develops an iterative gradient-steering" optimization framework. A network utility maximization problem is converted to a transmission scheduling problem that maximizes an approximated utility in its gradient direction, coupled with a steering vector update that continuously updates the approximated utility and its instantaneous gradient direction. The research then extends the framework to utility maximization for a network with multiple multicast sessions. Finally, the research develops distributed algorithms to optimize a global utility of a large scale network using local controllers. In addition to the planned research, the investigators will also try to extend the algorithm to ad hoc networks with time varying channels where utility maximization requires efficient exploitation of the channel diversity gain.