TMA from bearings and multipath time delays

A novel approach for target motion analysis (TMA), which uses conventional passive bearing together with multipath time-delay measurements is examined. This so-called "Multipath TMA" offers two tactical advantages over the classical bearings-only TMA: no requirement for any ownship maneuver, and a good performance in terms of estimation error achieved in a shorter time. Both known and unknown multipath cases are addressed. Finally, Monte-Carlo simulations and at-sea trials demonstrate the practical efficiency of such a multipath TMA.     

Discrete-time observability and estimability analysis for bearings-only target motion analysis

Observability in the context of bearings-only tracking (BOT) is still the subject of important literature. Different from previous approaches, where continuous-time analysis was considered, our approach relies on discrete-time analysis. It is then shown that this allows us to use directly and efficiently the simple formalisms of linear algebra. Using the direct approach, observability analysis is essentially reduced to basic considerations about subspace dimensions. Even if this approach is conceptually quite direct, it becomes more and more complex as the source-encounter scenario complexity increases. For complex scenarios, the dual approach may present some advantages essentially due to the direct use of multilinear algebra. New results about BOT observability for maneuvering sources are thus obtained. Observability analysis is then extended to unknown instants of source velocity changes. Even if observability analysis provides thorough insights about the algebraic structure of the BOT problem, the optimization of the observer maneuvers is essentially a control problem. Basic algebraic considerations prove that a relevant cost functional for this control problem is the determinant of the Fisher information matrix (FIM). So, a large part of this work is devoted to the analysis of this cost functional. Using multilinear algebra, general approximations of this functional are given. In order to involve only directly estimable parameters, the source bearing-rates are examined. Using these approximations, a general framework for optimizing the observer trajectory is derived which allow us to approximate the optimal sequence of controls. It is worth stressing that our approach does not require the knowledge of the source trajectory parameters and is still valid for a maneuvering source.     

Frequency line tracking using HMM-based schemes [passive sonar]

This paper deals with fluctuating line tracking, present on the so-called lofargram (low frequency analysis and recording) encountered in any passive sonar system. Considering such a line as a random walk modeled by a first-order Markov chain, we have recourse to the hidden Markov models (HMMs) arsenal. More precisely, we propose to track a frequency line with Viterbi and forward-backward algorithms. The originality of this work comes from the fact that a "probabilistic integration of the spectral power" approach allows us to construct a signal-to-noise (SNR)-knowledge-free method. Intensive simulations reveal no loss of performance.     

Estimation and classification of FM signals using time frequency transforms

This work deals with electronic warfare: we have to process an intercepted radar-emitted signal assumed to be frequency-modulated (FM). A battery of parametric models being available, we face the two main problems: the parameter estimation, given a model, and the classification, given the battery. For the former, we propose the classical maximum likelihood estimator (MLE) and an original numerical scheme to reach it. For the latter, we construct a test based on the estimated variances cross-ratios. Both perform very efficiently with respect to (w.r.t.) the Cramer-Rao lower bound (CRLB) and the rate of correct classification, respectively, for low signal-to-noise ratios (SNRs).     

Track formation with bearing and frequency measurements in clutter

Target motion analysis from a narrowband passive sonar that yields bearing and frequency measurements in the presence of false detections (clutter) in a low-SNR (low signal-to-noise ratio) environment is discussed. The likelihood function used to compute the maximum likelihood estimation of the track parameters (localization and frequency) incorporates the false alarms via the probabilistic data association technique. The Cramer-Rao lower bound is calculated and results obtained from simulations are shown to be compatible with it. A test of track acceptance is also presented     

A formulation of multitarget tracking as an incomplete data problem

Traditional multihypothesis tracking methods rely upon an enumeration of all the assignments of measurements to tracks. Pruning and gating are used to retain only the most likely hypotheses in order to drastically limit the set of feasible associations. The main risk is to eliminate correct measurement sequences. The probabilistic multiple hypothesis tracking (PMHT) method has been developed by Streit and Luginbuhl in order to reduce the drawbacks of "strong" assignments. The PMHT method is presented in a general mixture densities perspective. The Expectation-Maximization (EM) algorithm is the basic ingredient for estimating mixture parameters. This approach is then extended and applied to multitarget tracking for nonlinear measurement models in the passive sonar perspective.     

Observability in passive target motion analysis

The nature of observability of nonlinear systems encountered in passive target motion analysis (TMA) is carefully examined. The approach proposed here is based upon a well-chosen criterion which allows us to answer the major observability questions