Directed subspace search ML-PDA with application to active sonar tracking

Abstract

The Maximum Likelihood Probabilistic Data Association (ML-PDA) tracking algorithm is effective in tracking Very Low Observable targets (i.e., very low signal-to-noise ratio (SNR) targets in a high false alarm environment). However, the computational complexity associated with obtaining the track estimate in many cases has precluded its use in real-time scenarios. Previous ML-PDA implementations used a multi-pass grid search to find the track estimate. In this paper, two alternate methods for finding the track estimate are presented -a genetic search and a newly developed directed subspace (DSS) search algorithm. Each algorithm is tested using active sonar scenarios in which an autonomous underwater vehicle searches for and tracks a target. Within each scenario, the problem parameters are varied to illustrate the relative performance of each search technique. Both the DSS search and the genetic algorithm are shown to be an order of magnitude more computationally efficient than the multi-pass grid search, making possible real-time implementation. In addition, the DSS search is shown to be the most effective technique at tracking a target at the lowest SNR levels -reliable tracking down to 5 dB (post-processing SNR in a resolution cell) using a 5-frame sliding window is demonstrated, this being 6 dB better than the multi-pass grid search.

Integrated track initialization and maintenance in heavy clutter using probabilistic data association

“…Three methods have been used to compute the global maximum of JLLR: multi-pass grid (MPG) search [3,4,9], genetic search (GS) [24] and directed subspace search (DSS) [25]. It has been shown that DSS is the most effective technique when tracking a target at low SNR levels (6 dB) and more computationally efficient than the MPG search [25]. But there still is no guarantee that the resulting JLLR maximum is associated with the targets.…”

Tracking the Tracker from its Passive Sonar ML-PDA Estimates

“…In [7] ML-PDA was applied to the problem of low-observable target estimation using electro-optical sensors; also a slidingwindow batch approach for ML-PDA estimation was derived, capable of dealing with temporary disappearance of targets and/or targets with velocities changing over time. ML-PDA was also successfully applied to active sonar tracking in [6], where also an efficient computation of the ML estimate, namely, directed subspace search (DSS), was derived. The use of ML-PDA for early track detection with a radar is discussed in [2].…”

Noncontact Wideband Sonar for Human Activity Detection and Classification

“…The transition between the states depends on the PRF, and on the motion. Each "legal" transition between states at time instance m, can be defined by a branch matrix between objects k, and l, at instance time m can be defined similar to [33], to maximize the likelihood ratio of…”

“…Solution to the criterion in (9) is complex [33]. A sub-optimal solution, without the need to know the exact probability function of the object properties like in [36], would be to associate to each target or clutter its related objects separately based on the a-priori knowledge about the target properties, and then, in a later processing stage, exclude estimation errors from the targets estimations.…”