Optimality Self Online Monitoring (OSOM) for Performance Evaluation and Adaptive Sensor Fusion

Abstract

The performance of a tracking filter can be evaluated in terms of the filter's optimality conditions. Testing for optimality is necessary because the estimation error covariance as provided by the filter is not a reliable indicator of performance, which is known to be "optimistic" (inconsistent) particularly when there are model mismatches and target maneuvers. The conventional root-mean square (RMS) error value and its variants are widely used for performance evaluation in simulation and testing but it is not feasible for real-time operations where the ground truth is hardly available. One approach for real-time reliability assessment is optimality self online monitoring (OSOM) investigated in this paper. Statistical tests for optimality conditions are formulated. Simulation examples are presented to illustrate their possible use in evaluation and adaptation.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jul 01, 2008
Accession Number
ADA520525

Entities

People

  • Chun Yang
  • Erik Blasch
  • Ivan Kadar

Organizations

  • Air Force Research Laboratory

Tags

Communities of Interest

  • Sensors

DTIC Thesaurus Topics

  • Algorithms
  • Data Science
  • Detectors
  • Estimators
  • Information Processing
  • Information Science
  • Kalman Filters
  • Knowledge Management
  • Monitoring
  • Resource Management
  • Sensor Fusion
  • Statistical Algorithms
  • Statistics
  • Steady State
  • Target Recognition
  • Target Tracking
  • Test And Evaluation

Fields of Study

  • Engineering

Readers

  • Computational Modeling and Simulation
  • Inertial Navigation Systems.
  • Operations Research