Structure-Aware Reliability Analysis of State Estimators in Large-Scale Sensor Systems
Grant
Overview
Affiliation
View All
Overview
abstract
The proposed research aims to develop efficient methods to measure the reliability of state estimators in large-scale sensor systems by exploiting the inherent structures of these systems and verify the effectiveness of the developed methods in practice. Various infrastructures, which are monitored and controlled using large-scale sensor systems, form the backbone of Qatarâ s economic and environmental development. The reliability of estimations made by these sensor systems is crucial to smooth operation of these infrastructures. Yet, no efficient method to calculate this reliability in large-scale sensor systems exists. The PIs believe that when certain types of inherent system structure are exploited, computational efficiency of calculating estimation reliability can be improved by orders of magnitude. Following successful preliminary results, the PIs propose substantial new research to relax the restrictive conditions under which the current methods work. If successful, this research will result in exact algorithms based on 0-1 mixed integer programming (MIP) decomposition, convex approximation algorithms, and 0-1 MIP/matroid decomposition hybrid algorithms to calculate estimation reliability. It will also produce enhanced methods to measure stochastic estimation reliability of a sensor system given the failure probabilities of the individual sensors. The practical performance of the developed methods will be verified on a real wireless sensor network used for environmental and ecological monitoring, which will be set up by the research team as a pilot study.
