NeTS: Small: Distributed Wireless Networking with An Enhanced Physical-Link Layer Interface

未来无线系统的发展需要一种同时解决通信效率和系统模块化的网络体系结构。由于现有网络体系结构的低效，现有网络大多采用仅在最后一跳使用无线通信的结构。本项目的目标是开发一种增强的物理链路层接口，以解决特殊结构网络中无线部分的体系结构低效问题，主要针对分布式通信和组网模型。通过为链路层用户配备多种传输选项，新接口可实现链路层的高级通信适配，从而利用功率、速率和天线调整等无线功能。该项目代表了对分布式链路层联网的第一次系统调查，该分布式链路层联网利用每个链路层用户的多个传输选项。该项目将为信息和网络理论的整合建立一座重要的桥梁，并将成为寻求将其对有线系统运行的干扰降至最低的无线体系结构开发的范例。信息理论和网络理论的统一也将影响高等教育学科的教学方式，并吸引更多有才华的学生进入这个对国家具有重大意义的领域。该项目分为四个步骤。在步骤1中，重点研究具有同构用户和饱和消息队列的典型分布式网络场景。基于随机近似模型，将开发一个媒体访问控制(MAC)框架，以实现对分布式信道共享问题的有效描述。分布式算法将被开发来确保关于特定网络效用函数的最大化的渐近最优或接近最优的性能。在步骤II中，重点是将MAC框架扩展到包含不同优先级用户组的分布式网络，以及要求次用户不干扰主用户通信的分布式认知网络。在第三步中，重点是通过将指数退避方法结合到MAC框架中来开发快速自适应MAC算法。在第四步中，重点是使用计算机仿真来研究MAC算法在实际场景中的性能。

The development of future wireless systems requires a network architecture that simultaneously addresses communication efficiency and system modularity. Due to the inefficiency of current network architectures, most existing networks adopt the structure of using wireless communication only at the last hop. The goal of this project is to develop an enhanced physical-link layer interface to solve the architectural inefficiency problem in wireless part of the special-structured networks, mainly for the distributed communication and networking model. By equipping a link layer user with multiple transmission options, the new interface enables advanced communication adaptation at the link layer that exploits wireless capabilities such as power, rate, and antenna adjustments. This project represents the first systematic investigation of distributed link layer networking that exploits multiple transmission options at each link layer user. The project will establish an important bridge toward the integration of information and network theories, and will serve as an example wireless architecture development that seeks to minimize its disturbance to the operation of wireline systems. Unification of information theory and network theory will also influence the way subjects are taught in Higher Education and attract more talented students to this field of acute national importance. The project is organized into four steps. In Step I, the focus is to investigate typical distributed networking scenarios with homogeneous users and saturated message queues. Based on a stochastic approximation model, a medium access control (MAC) framework will be developed to enable effective formulation of distributed channel sharing problems. Distributed algorithms will be developed to ensure asymptotic optimal or near optimal performance with respect to the maximization of particular network utility functions. In Step II, the focus is to extend the MAC framework to distributed networks that contain user groups with different priorities, as well as to distributed cognitive networks where secondary users are required not to disturb the communications of the primary users. In Step III, the focus is to develop fast adaptive MAC algorithms by incorporating the exponential backing-off approach into the MAC framework. In Step IV, the focus is to use computer simulations to investigate performance of the MAC algorithms in practical scenarios.