ACOUSTIC LOCALIZATION FROM BEARING AND FREQUENCY AT SPATIALLY DISTRIBUTED SENSORS

PDF

Authors
  1. Jenkin, M.
  2. Milios, E.
Corporate Authors
Defence Research Establishment Atlantic, Dartmouth NS (CAN);York Univ, North York ONT (CAN) Dept of Computer Science
Abstract
The traditional approach for solving the problem of sonar information management using passive sensors is a bottom-up process, in which signal tracking (temporal association) is performed first, followed by intrasensor and intersensor signal association, and finally source localization and tracking. The philosophy of the work described in this report is that signal tracking and association can be performed more robustly if coupled more tightly with source localization and tracking. Errors in signal tracking and association can be identified because they result in "unreasonable" source locations and tracks. This report specifically addresses the problem of localizing multiple targets that follow piecewise linear paths using a field of passive sonobuoys. The report first presents two robust statistics-based techniques for simultaneously solving the signal association and target localization problems, assuming that target manoeuvres have been correctly identified. The first method, robust outlier detection and elimination, is based on the identification and elimination of outlier data points during least-squares fitting. The second method, mixture models, assumes that the measurements originate from one or more data sources characterized by class probabilities. Both techniques can extract multiple sources and their tracks from bearing/frequency measurements that have not been associated with individual sources. TRUNCATED
Keywords
Acoustic localization;Signal association;Robust outlier detection;Mixture models
Report Number
DREA-CR-98-422 — Contractor Report
Date of publication
01 Mar 1997
Number of Pages
41
DSTKIM No
98-01875
CANDIS No
508512
Format(s):
Hardcopy;Document Image stored on Optical Disk

Permanent link

Document 1 of 1

Date modified: