Information Measures for Multisensor Systems

Abstract

The purpose of this report is to demonstrate the utility of an information-theoretic approach to next generation chemical detection. Recent research at the Naval Research Laboratory (NRL) has yielded probabilistic models for spectral data that enable the computation of information measures such as entropy and divergence, with the goal of developing feature sets to increase the sensitivity and selectivity of multivariate chemical sensors of several modalities. Results are presented for several types of spectral data in multisensor systems, as well as strategies for using information measures with other data sources. Binary, univariate, and multivariate sensors can all be modeled from an information-theoretic perspective, making it well-suited for the challenges of next generation chemical detection.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Dec 11, 2013
Accession Number
ADA591226

Entities

People

  • Christian P. Minor
  • Joseph C. Gezo
  • Kevin Johnson

Organizations

  • United States Naval Research Laboratory

Tags

Communities of Interest

  • Materials and Manufacturing Processes
  • Sensors

DTIC Thesaurus Topics

  • Algorithms
  • Chemical Compounds
  • Chemical Detection
  • Chemistry
  • Data Sets
  • Detection
  • Detectors
  • Information Theory
  • Infrared Spectra
  • Infrared Spectroscopy
  • Mass Spectra
  • Mass Spectrometry
  • Probability
  • Probability Distributions
  • Spectra
  • Spectrometry
  • Spectroscopy

Readers

  • Computational Modeling and Simulation
  • Regression Analysis.
  • Sensor Fusion and Tracking Systems.