CSR: Small: Collaborative Research: Integrated Control of Fidelity and Real-Time Performance in Networked Sensing Systems

网络感知系统(NSS)通过将感知、计算、网络和物理过程紧密集成在一起，与物理世界交互。NSS系统正被部署在许多关键应用中，如安全、民用基础设施、医疗保健、制造和交通。这些应用程序通常会施加严格的性能要求。NSS系统的许多计算和通信任务必须在一定的时间限制内完成，以避免不受欢迎的甚至灾难性的后果--这一特性被称为实时保证。此外，即使在输入和/或内部系统组件出现噪声和错误时，关键任务NSS系统也必须将其性能保持在可接受的水平--这是一种称为保真度的特性。本研究旨在设计一种新颖的NSS控制框架，将数据融合、校准和实时性能控制集成到一个平衡保真度和实时保证要求的解决方案中。预期成果包括：1)用于保真度和实时保证的统一多层性能控制框架；2)通过新的模型和融合校准算法改进保真度保证；3)保真度感知实时性能控制机制；以及4)几个关键任务领域的集成保真度和实时保证，包括实时火山监测和断层成像以及高保真特别监视。该项目对未来多个应用领域的NSS系统具有广泛的意义，这些应用领域需要在严格的时间约束下高保真地处理动态和复杂的物理信息。教育和外联活动包括在两个新的研究生课程中引入国家支持系统，并在项目过程中开发软件、试验床和实验室，以及招募妇女和少数族裔学生参加该项目。

Networked Sensing Systems (NSSs) interact with the physical world by tightly integrating sensing, computation, networking, and physical processes. NSS systems are being deployed for numerous critical applications such as security, civil infrastructure, healthcare, manufacturing, and transportation. These applications often impose stringent performance requirements. Many computation and communication tasks of NSS systems must be finished within certain timing constraints to avoid undesirable or even catastrophic consequences -- a property known as real-time assurance. In addition, mission-critical NSS systems must maintain their performance at an acceptable level even when noises and errors occur in input and/or internal system components -- a property known as fidelity. This research is to design a novel NSS control framework that integrates data fusion, calibration, and real-time performance control into a solution that balances requirements for fidelity and real-time assurance. The expected outcomes include: 1) a unified multi-tier performance control framework for both fidelity and real-time assurance; 2) improvements in fidelity assurance through novel model and fusion calibration algorithms; 3) a fidelity-aware real-time performance control mechanism; and 4) integrated fidelity and real-time assurance for several mission-critical domains including real-time volcano monitoring and tomography, and high-fidelity ad hoc surveillance. This project has broad implications for future NSS systems in multiple application domains that require high-fidelity processing of dynamic and complex physical information within stringent timeliness constraints. Educational and outreach activities include introduction of NSS systems into two new graduate courses with software, testbed, and labs developed in the course of this project, and recruitment of women and minority students for participation in the project.