Methodology for Monitoring Complex Systems Using Large Scale Sensor Networks

这笔赠款提供资金，以建立一个统一的框架，用于管理大型嵌入式传感器网络，用于制造企业和在随机环境中运行的复杂系统的健康监测。 该项目将侧重于自适应传感和传感器通信的分布式推理和决策领域的基础研究问题。具体而言，它将处理以下问题：1）开发用于管理大量联网传感器的理论基础，这些联网传感器共同形成“智能”感测网格，能够进行分布式推断，并保证数据解释的一致性，以及2）部署可联网的结构配置和参数设计，微型传感器，以物理实现拟议的“智能”传感网格使用一种新的“代理芯片”（AOC）固件。如果成功，这项研究的结果将导致在两个领域的改进。首先，该研究将导致一个新的理论和定量的指导方针的架构设计和部署的网络传感器，其特点是能够执行事件驱动，动态推理和分布式决策在本地传感器的水平。其次，它将证明，一个新的传感器网络架构，将物理体现，台架测试，并验证了大量的传感器原型的基础上，最先进的小型化技术的可行性。这些硬件和软件将适用于广泛的应用领域，包括制造、运输、医疗保健、医疗设备、环境控制和国土安全。这项研究的更广泛的影响将是弥合决策理论，贝叶斯网络和嵌入式传感芯片设计在学术界和制造环境中的实际应用的基础研究之间的差距。在教育方面，拟议的活动将有助于对研究生和本科生进行大规模传感器网络管理领域的多学科培训。

This grant provides funding to establish a unified framework for the management of large-scale, embedded sensor networks for the health monitoring of manufacturing enterprises and complex systems operating in stochastic environments. The project will focus on fundamental research issues in the areas of distributed inferencing and decision-making for adaptive sensing and sensor communication. Specifically, it will address the following issues: 1) development of a theoretical basis for managing large number of networked sensors that collectively form an "intelligent" sensing grid, capable of making distributed inferences with guarantees on consistency of data interpretation, and 2) deployment of architectural configuration and parametric design of networkable, miniaturized sensors to physically realize the proposed "intelligent" sensing grid using a novel "agent-on-a-chip" (AOC) firmware.If successful, the results of this research will lead to improvements in two areas. First, the research will lead to a new theory and quantitative guidelines for the architectural design and deployment of networked sensors that are characterized by the ability to perform event-driven, dynamic inferencing and distributed decision-making at the local sensor level. Secondly, it will demonstrate, the feasibility of a new sensor network architecture that will physically embody, bench test, and validate a large number of sensors prototyped based on the state-of-the-art miniaturization technologies. Such hardware and software will be applicable in a wide range of application areas, including manufacturing, transportation, health care, medical devices, environmental control, and homeland security. The broader impact of this research will be in bridging the gap between fundamental research on decision theory, Bayesian networks, and embedded sensing chip design in the academia and real-world applications in a manufacturing environment. On the educational front, the proposed activity will contribute to multidisciplinary training of graduate and undergraduate students in the area of management of large-scale sensor networks.