Sound Speed Estimation as a Means of Improving Target Tracking Performance,
Abstract
THIS PAPER ADDRESSES THE ISSUE OF TARGET TRACKING WHEN CONFRONTED WITH A SET OF SOUND SPEED PARAMETERS THAT ARE PARTIALLY OR COMPLETELY UNKNOWN. It explores the case where these parameters are augmented to the target state in an extended Kalman filter. The filter processes measurements of sound time-of-arrival difference and Doppler difference from a set of spatially displaced sensors. For scenarios involving up to three sensors it has been found that biased target position estimates and marginal system observability occurs. This is readily verified by propagating the engine values of the information matrix in time. Using this as an analysis tool, a number of geometrical sensor configurations are analyzed. In general, it is found that with three sensors, the system is, at best, marginally observable for any geometry. However, when using four and more sensors, system observability and estimation performance are markedly improved when two of the sound speed filter parameters are specified to within a close tolerance of their actual values. When attempting to estimate all of the sound speeds (or for that matter (n-1)) sound speeds, n = number of sensors), it is again noted, as in the three-sensor case, that system observability and estimation performance become degraded. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1979
- Accession Number
- ADA086603
Entities
People
- D. L. Alspach
- G. L. Mohnkehn
- R. N. Lobbia