From Medium Access to Physical Layer: An Integrated DSP Framework for Wireless Packet Networks

第一代和第二代无线网络在提供全球覆盖的语音服务方面取得了显著的成功。基于这一成功，第三代和第四代网络将更加强调灵活的数据服务，而不是语音服务。这种基础设施将能够实现可能尚未想象到的先进和多样化的无线服务。实现高速无线数据网络最棘手的技术挑战之一是在无线信道上多路复用可变比特率数据用户。像TDMA、FDMA甚至CDMA这样的经典解决方案是不合适的，因为它们坚持面向连接，而不是基于包的无连接哲学。另一方面，随机访问技术提供了所需的灵活性，但在繁忙的流量下遭受严重的吞吐量/延迟损失。为了处理预期的多种业务组合，必须开发新的介质接入（MAC）方案。正如NSF 99-68倡议所强调的那样，越来越多的人认为，这种新颖的解决方案需要打破传统的分层网络设计方法，转向跨多层的集成设计。本研究的目标是一种跨越介质访问层和物理层的无线分组网络设计的集成方法。从最近引入的MAC协议（称为网络分集多址NDMA）开始，该协议利用MAC级网络资源在物理层和MAC层之间创建协同作用，研究人员研究分组无线网络中的联合碰撞分辨率，盲波束形成和功率控制，目标是展示比传统分层设计更大的吞吐量延迟性能增益。这项研究涉及先进的信号处理、因子分析、优化和排队工具。为了将NDMA概念发展成为一个完整的最先进的物理和数据链路层解决方案，正在研究以下具体问题：(i)实际信噪比的NDMA性能分析和频率选择通道的系统设计；（ii）开发盲和半盲NDMA碰撞解决方法及相关性能分析；（iii）系统集成问题：分组无线网络中的联合盲波束形成、碰撞分辨率、功率控制和ARQ设计。该项目还包括一个综合教育组成部分，促进研究生和本科生在无线通信和网络方面的研究。

First and second generation wireless networks have had a remarkable success in providing voice services with global coverage. Building on this success, third and fourth generation networks are envisioned with more emphasis on flexible data rather than voice services. Such an infrastructure will enable the implementation of advanced and diverse wireless services which are perhaps not yet imagined. One of the thorniest technical challenges in implementing high speed wireless data networks is that of multiplexing variable-bit-rate data users over the wireless channel. Classical solutions like TDMA, FDMA, or even CDMA are not appropriate, since they adhere to a connection-oriented, rather than packet-based, connectionless philosophy. Random access techniques on the other side offer the required flexibility, but suffer from severe throughput/delay penalties under heavy traffic. In order to handle the anticipated diverse traffic mix, novel Medium Access (MAC) schemes have to be developed. As the NSF 99-68 initiative underlines, there is growing consensus that such novel solutions require breaking away from the traditional layered approach to network design, moving towards integrated design across multiple layers.The goal of this research is an integrated approach to wireless packet network design that cuts across the medium access and physical layers. Starting from a recently introduced MAC protocol (termed NDMA for Network Diversity Multiple Access) which utilizes MAC-level network resources to create a synergy between the physical and MAC layers, the investigators study joint collision resolution, blind beamforming, and power control in packet radio networks, with the goal of demonstrating substantial throughput-delay performance gains over traditional layered designs. This research involves advanced signal processing, factor analysis, optimization, and queueing tools. Aimed at developing the NDMA concept into a complete state-of-art physical and data link layer solution, the following specific issues are under investigation: (i)NDMA performance analysis for realistic SNR, and system design for frequency selective channels; (ii)development of blind and semi-blind NDMA collision resolution methods and associated performance analysis; and (iii)system integration issues: joint blind beamforming, collision resolution, power control, and ARQ design in packet radio networks. The project also includes an integrated educational component promoting graduate and undergraduate research in wireless communications and networking.