Loading [MathJax]/extensions/MathZoom.js
Target Tracking With Particle Filters Under Signal Propagation Delays | IEEE Journals & Magazine | IEEE Xplore

Target Tracking With Particle Filters Under Signal Propagation Delays


Abstract:

Signal propagation delays are hardly a problem for target tracking with standard sensors such as radar and vision due to the fact that the speed of light is much higher t...Show More

Abstract:

Signal propagation delays are hardly a problem for target tracking with standard sensors such as radar and vision due to the fact that the speed of light is much higher than the speed of the target. This contribution studies the case where the ratio of the target and the propagation speed is not negligible, as in the case of sensor networks with microphones, geophones or sonars for instance, where the signal speed in air, ground and water causes a state dependent and stochastic delay of the observations. The proposed approach utilizes an augmentation of the state vector with the propagation delay in a particle filtering framework to compensate for the negative effects of the delays. The model of the physics rules governing the propagation delays is used in interaction with the target motion model to yield an iterative prediction update step in the particle filter which is called the propagation delayed measurement particle filter (PDM-PF). The performance of PDM-PF is illustrated in a challenging target tracking scenario by making comparisons to alternative particle filters that can be used in similar cases.
Published in: IEEE Transactions on Signal Processing ( Volume: 59, Issue: 6, June 2011)
Page(s): 2485 - 2495
Date of Publication: 03 March 2011

ISSN Information:


I. Introduction

The conventional sensors such as radar, vision (EOR, IR), etc. used in target tracking [1]–[3] generally observe emitted (passive sensors) or reflected (active sensors) energy from the target. The observation delay in these sensors is negligible since the speed of light (electromagnetic waves) is much larger than the speed of the target. However, one trend in sensor networks is to use standard low cost sensors as microphones and geophones on land and sonar in water. The assumption of negligible target speed compared to the speed of the media cannot always be made here. In a general scenario where the target moves swiftly, even if the sensor is stationary and collects uniformly sampled (in time) measurements of the target, the actual time instants that the target is observed are in fact nonuniform (in time) due to the propagation delays and this leads to unexpected errors in the estimation algorithms.

References

References is not available for this document.