CAREER: Information-driven distributed coordination of mobile sensor networks in dynamic scenarios

职业：动态场景下信息驱动的移动传感器网络分布式协调

• 批准号: 0830601
• 负责人: Jorge Cortes
• 金额: $ 28.37万
• 依托单位: University of California-San Diego
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2008
• 资助国家: 美国
• 起止时间: 2008-01-01 至 2012-02-29
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0830601&HistoricalAwards=false
• 关键词: CAREER; Information; driven; distributed; coordination

Project SummaryThe emergence of low-cost, highly-autonomous vehicles equipped with control, communication, sensingand computing capabilities is paving the way for the deployment of mobile sensor networks in a wide rangeof applications. Examples include environmental monitoring, oceanographic research, high-stress, rapid de-ployment operations, and health monitoring of civil infrastructure. In these envisioned applications, manycritical processes occur at temporal and spatial scales that cannot be effectively sampled with current ap-proaches. Mobile sensor networks hold the promise to provide the rich, in-situ spatio-temporal data needed torevolutionize the detection, estimation, and monitoring of dynamic natural phenomena. Controlled mobilityintegrated with distributed data fusion capabilities will enable sensor networks to provide broad spatial cov-erage, react to short-lived events in real time, and track key processes that occur away from fixed sites.The state of the art in distributed data fusion only considers static networks, and therefore is not directlyapplicable to ad-hoc, dynamically changing mobile networks. The state of the art in motion coordinationof networked systems has only developed centralized approaches to estimation and signal processing. As aresult of these limitations, current mobile networks are too rigid to cope with the small-scale features andthe rapid evolution characteristic of many key physical processes.Intellectual Merit. The major objective of this project is the synthesis of scalable coordination algorithmsfor mobile networks performing spatially-distributed sensing tasks. Distributed strategies that maximize theinformation content of collected data will allow future sensor networks to adapt to changing conditions in arapid, autonomous and optimal fashion. To make this vision a reality, this project will address the distributed,in-situ aggregation of data collected by mobile networks in dynamic scenarios, and the information-driven,scalable coordination of the network mobility to optimally perform the required sensing tasks.The research plan will adopt an ambitious integrative approach composed of three thrusts: (i) a sound,unifying framework where different cooperative strategies can be rigorously formalized and compared. Thiseffort will facilitate the modular design of cooperative strategies for complex sensing tasks via the combinationof simpler algorithms performing more basic objectives; (ii) system-theoretic tools to evaluate the correctness,robustness and scalability properties of coordination algorithms. To assess the optimal trade-offs betweenperformance and energy allocation in combined communication, motion, and sensing scenarios, this researchwill evaluate complexity measures for cooperative strategies and spatially-distributed tasks; and (iii) novel,systematic design methodologies that allow to break down global sensing tasks into local objectives forindividual agents. This effort seeks to synthesize fault tolerant, scalable algorithms that sit at the limits onthe achievable performance, operational time and energy consumption of mobile networks conducting datafusion and estimation tasks. The innovative technical approach builds on a set of very promising results bythe PI and collaborators, hinging upon disciplines such as cooperative and topology control, automata andhybrid systems theory, robotics, wireless communications, and operations research.Broader Impacts. The techniques developed in this work will help design autonomous and efficient mobilenetworks performing critical tasks in homeland security, industrial processes, health care and the environment.The proposed research will lead to crosscutting and synergistic technologies applicable to a wide rangeof scenarios where real-time information gathering and data exploitation are important. The results willbe transferred to oceanographic and disaster management applications in collaboration with the MontereyBay Aquarium Research Institute, and NASA Ames, respectively.The proposed educational activities are integrated into the research plan and consist of the followinginitiatives: (i) involvement of undergraduate students in research via design projects, summer internshipsand engineering research demonstrations in a newly-created laboratory environment; (ii) development of anundergraduate course on \Motion Coordination" and a graduate course on \Cooperative Mobile Networks";(iii) offering of a course on control and robotics in the California State Summer School for Mathematics andScience for high-school students, and expository and research talks at community colleges near UCSC. Re-search and educational materials will be made available to high-school teachers, the scientific community andthe general public via an interactive website. Regular activities for broad dissemination (journal publications,conference presentations, lecture notes) will also be pursued. The evaluation of the educational activities,based on the educational process and the students' outcomes, will be done in collaboration with the UCSCCenter for Teaching Excellence. Special attention will be paid to integrate inclusive teaching practices intothe daily educational activity to address retention issues concerning underrepresented students.

项目概述具有控制、通信、传感和计算能力的低成本、高度自主的车辆的出现，为移动传感器网络在广泛的应用中的部署铺平了道路。例如，环境监测、海洋研究、高压力、快速部署作业以及民用基础设施的健康监测。在这些设想的应用中，许多关键过程发生在时间和空间尺度上，而目前的方法无法有效地对这些过程进行采样。移动传感器网络有望提供丰富的、现场的时空数据，以彻底改变对动态自然现象的检测、评估和监测。受控移动性与分布式数据融合能力相结合，将使传感器网络能够提供广泛的空间覆盖，对短暂的事件做出实时反应，并跟踪发生在固定地点以外的关键过程。分布式数据融合的现状只考虑静态网络，因此不能直接适用于自组织、动态变化的移动网络。网络系统的运动协调的最新水平只发展了用于估计和信号处理的集中式方法。由于这些限制，目前的移动网络过于僵化，无法适应许多关键物理过程的小规模特征和快速演变特征。该项目的主要目标是合成可扩展的协调算法，用于执行空间分布的感知任务的移动网络。将收集到的数据的信息量最大化的分布式策略将允许未来的传感器网络以快速、自主和优化的方式适应不断变化的条件。为了实现这一愿景，该项目将解决移动网络在动态场景中收集的数据的分布式、就地聚合，以及网络移动性的信息驱动、可扩展的协调，以最佳地执行所需的感知任务。研究计划将采用一种雄心勃勃的综合方法，由三个方面组成：(I)健全、统一的框架，其中不同的合作策略可以被严格地形式化和比较。这项工作将通过组合执行更基本目标的更简单的算法来促进复杂传感任务的协作策略的模块化设计；(Ii)评估协调算法的正确性、稳健性和可扩展性的系统论工具。为了评估在组合通信、运动和感知场景中性能和能量分配之间的最佳权衡，本研究将评估合作策略和空间分布任务的复杂性度量；以及(Iii)允许将全局感知任务分解为单个代理的局部目标的新颖、系统的设计方法。这项工作旨在合成容错、可扩展的算法，这些算法位于移动网络进行数据传输和估计任务的可实现性能、运行时间和能量消耗的极限。这一创新的技术方法建立在PI和合作者的一系列非常有希望的成果的基础上，依赖于协作和拓扑控制、自动机和混合系统理论、机器人学、无线通信和运筹学等学科。这项工作开发的技术将有助于设计自主和高效的移动网络，在国土安全、工业流程、医疗保健和环境中执行关键任务。拟议的研究将导致可应用于实时信息收集和数据开发的广泛场景的横切和协同技术。结果将分别与蒙特利湾水族馆研究所和NASA Ames合作，转移到海洋和灾害管理应用中。拟议的教育活动被整合到研究计划中，包括以下举措：(I)通过设计项目、暑期实习和在新创建的实验室环境中进行工程研究演示，让本科生参与研究；(Ii)开发关于运动协调的本科生课程和关于合作移动网络的研究生课程；(Iii)在加州州立数学与科学暑期学校为高中生提供关于控制和机器人的课程，以及在南加州大学附近的社区大学进行暴露和研究讲座。研究和教育材料将通过一个互动网站向高中教师、科学界和普通公众提供。还将开展广泛传播的定期活动(期刊出版物、会议报告、讲座讲稿)。对教育活动的评估将基于教育过程和学生的结果，将与加州大学洛杉矶分校卓越教学中心合作进行。将特别注意将兼容并包的教学做法纳入日常教育活动，以解决有关代表性不足学生的留住问题。