Non-Ergodic Wireless Sensing: Fundamental Tradeoffs and Optimal Transmission Schemes

作为未来信息基础设施的重要组成部分，无线传感系统激发了社会各界的巨大研究兴趣。这些努力对于建立物理世界和网络世界之间的桥梁至关重要，在网络世界中，可以收集和处理物理系统参数，如电网中的电压/电流水平、农田上的温度/湿度水平以及桥梁上的压力水平，以便进行自动网络控制。然而，如何找到在这样的传感器网络上传递信息的最佳方式仍然是一个悬而未决的问题，因为对于端到端的信息收集、传输和处理缺乏明确的最优标准。在该项目中，PI首先基于新的网络级多路复用分集折衷分析提出了新的优化准则，然后研究了有效的信息传输和处理方案，特别考虑了恶意应用环境下网络的非遍历性。拟议中的研究将对我们的社会产生巨大的多方面影响。传感系统的设计成果将使各种关键应用成为可能，如复杂的网络物理系统监测和大规模实时环境监测。为传感器网络开发的基本性能分析也可以应用于传感应用以外的其他类型的网络。拟议的计划将通过为学生提供培训、加强学校课程、定期举办研讨会、在项目网站上公布结果以及在技术会议上提交论文来扩大社会的教育范围。具体地说，在这个项目中，PI从解决无线传感系统设计中的一个基本问题开始：给定M个未知实体(源)，由N个无线传感节点组成的网络在功率、时间、频率和空间上使用一定的资源来感知，在可支持的M个、可实现的端到端可靠性和所需的网络资源之间将进行什么基本的权衡。非遍历随机通信信道和非遍历随机节点丢失被联合考虑。对相关和不相关信号源的情况都进行了深入的研究。受多天线传输系统中复接和分集折衷的启发，重新定义了复接和分集折衷的概念，将其作为分布式传感器网络环境下的最优准则。该项目的重点是量化不同感兴趣实体之间可实现的基本折衷，并提出有效的信息传输和处理方案以实现最优折衷曲线上的某些特定操作点，并特别考虑随机传感器网络在各种应用中的非遍历性质。可交付的内容包括基本理论、性能界限和最优感知方案。智能的优点在于对无线传感系统中的核心设计问题进行了系统的处理，该系统适用于具有非遍历网络随机性的应用。深入研究了基于估计/检测中断和估计/检测分集的变革性新概念，并将其推广到相关信源、多跳网络和多融合中心的情况。

As an essential element in the future information infrastructure, wireless sensing systems have motivated a huge amount of research interests in the community. Such efforts are critical to building the bridge between the physical world and the cyber world, where physical system parameters such as voltage/current levels in a power grid, temperature/humidity levels over a farming field, and stress levels across a bridge could be collected and processed for automatic cyber control. However, it is still an open problem on how to find the optimal way of conveying information over such sensor networks, since there is a lack of well-defined optimality criteria for end-to-end information collection, transmission, and processing. In this project, the PI first proposes new optimality criteria based on novel network-wise multiplexing diversity tradeoff analysis, then studies the efficient information transmission and processing schemes, with special considerations on network non-ergodicity in hostile application environments. The proposed research will have great multi-fold impacts on our society. The sensing system design results will enable various key applications such as complex cyber-physical system monitoring and large-scale real-time environment surveillance. The fundamental performance analysis developed for sensor networks could also be applied to other types of networks beyond the sensing applications. The proposed program will broaden the education scope in the society by providing training to students, enhancing school curriculum, holding regular seminars, posting results on the project website, and presenting papers in technical conferences. It will also emphasize the outreaches to women and under-represented groups.Specifically, in this program, the PI starts by addressing one fundamental problem in wireless sensing system design: Given M unknown entities (sources) to sense by a network of N wireless sensing nodes operating with certain resources in power, time, frequency, and space, what the fundamental tradeoffs will be among the supportable M, the achievable end-to-end reliability, and the required network resources. Non-ergodic random communication channels and non-ergodic random node losses are jointly considered. Both cases of correlated and uncorrelated sources are thoroughly studied. Motivated by the multiplexing and diversity tradeoff result in multi-antenna transmission systems, the concept of multiplexing and diversity tradeoff is redefined as the optimality criteria in the distributed sensor network context. The emphasis of this project is quantifying the fundamental achievable tradeoffs among various entities of interest, and proposing efficient information transmission and processing schemes to achieve some particular operation points on the optimal tradeoff curves, with special considerations over the non-ergodic natures of random sensor networks in various applications. The deliverables include basic theories, performance bounds, and optimal sensing schemes. The intellectual merits lie in the systematic treatment over core design issues in wireless sensing systems operating for applications with non-ergodic network randomness. Transformative new concepts based on estimation/detection outage and estimation/detection diversity are thoroughly investigated and generalized to cases involving correlated sources, multi-hop networks, and multiple fusion centers.