Towards Intelligent Dynamic Deployment of Mobile Sensors in Complex Resource-Bounded Environments

Ng, B M ; Hanley, W G

Lawrence Livermore National Laboratory

关键词: Decision Making;    Diseases;    Remote Sensing;    99 General And Miscellaneous//Mathematics, Computing, And Information Science;    Data Acquisition;   

DOI  :  10.2172/919953
RP-ID  :  UCRL-TR-231422
RP-ID  :  W-7405-ENG-48
RP-ID  :  919953

美国|英语

来源: UNT Digital Library

【 摘 要 】

Decision-making in the face of uncertainty requires an understanding of the probabilistic mechanisms that govern the complex behavior of these systems. This issue applies to many domains: financial investments, disease control, military planning and homeland security. In each of these areas, there is a practical need for efficient resource-bounded reasoning capabilities to support optimal decision-making. Specifically, given a highly complex system, with numerous random variables and their dynamic interactions, how do we monitor such a system and detect crucial events that might impact our decision making process? More importantly, how do we perform this reasoning efficiently--to an acceptable degree of accuracy in real time--when there are only limited computational power and sensory capabilities? These questions encapsulate nontrivial key issues faced by many high-profile Laboratory missions: the problem of efficient inference and dynamic sensor deployment for risk/uncertainty reduction. By leveraging solid ideas such as system decomposition into loosely coupled subsystems and smart resource allocation among these subsystems, we can parallelize inference and data acquisition for faster and improved computational performance. In this report, we propose technical approaches for developing algorithmic tools to enable future scientific and engineering endeavors to better achieve the optimal use of limited resources for maximal return of information on a complex system. The result of the proposed research effort will be an efficient reasoning framework that would enable mobile sensors to work collaboratively as teams of adaptive and responsive agents, whose joint goal is to gather useful information that would assist in the inference process.