Asma Rabaoui

Dirichlet Process Mixtures for Density Estimation in Dynamic Nonlinear Modeling: Application to GPS Positioning in Urban Canyons

In global positioning systems (GPS), classical localization algorithms assume, when the signal is received from the satellite in line-of-sight (LOS) environment, that the pseudorange error distribution is Gaussian. Such assumption is in some way very restrictive since a random error in the pseudorange measure with an unknown distribution form is always induced in constrained environments especially in urban canyons due to multipath/masking effects. In order to ensure high accuracy positioning, a good estimation of the observation error in these cases is required. To address this, an attractive flexible Bayesian nonparametric noise model based on Dirichlet process mixtures (DPM) is introduced. Since the considered positioning problem involves elements of non-Gaussianity and nonlinearity and besides, it should be processed on-line, the suitability of the proposed modeling scheme in a joint state/parameter estimation problem is handled by an efficient Rao-Blackwellized particle filter (RBPF). Our approach is illustrated on a data analysis task dealing with joint estimation of vehicles positions and pseudorange errors in a global navigation satellite system (GNSS)-based localization context where the GPS information may be inaccurate because of hard reception conditions.

Using robust features with multi-class SVMs to classify noisy sounds

In a sounds recognition system, the most encountered problem is the background noise that can be captured with the sounds to be identified. This paper describes work that has been performed to address this problem. First, the robustness to the environmental noise is investigated for specific kinds of acoustic representation. The representations considered are RASTA-PLP, J-RASTA and wavelets-based processing. Then, we propose to apply multi-class support vector machines (SVMs) as a discriminative framework in order to address audio classification. The experiments conducted on a multi- class problem show that this classifier clearly overperforms the conventional HMM-based system, and hence, we can efficiently address a sounds classification problem characterized by complex real-world datasets, even under important noise degradation conditions.

GNSS pseudorange error density tracking using Dirichlet Process Mixture

In satellite navigation system, classical localization algorithms assume that the observation noise is white-Gaussian. This assumption is not correct when the signal is reflected on the surrounding obstacles. That leads to a decrease of accuracy and of continuity of service. To enhance the localization performances, a better observation noise density can be use in an adapted filtering process. This article aims to show how the Dirich-let Process Mixture can be employed to track the observation density on-line. This sequential estimation solution is adapted when the noise is non-stationary. The approach will be tested under a simulation scenario with multiple propagation conditions. Then, this density modeling will be used in Rao-Blackwellised Particle Filter.

Studies on DPM for the density estimation of pseudorange noises and evaluations on real data

GNSS localization is accurate in clear environment where the pseudorange noise distributions are assumed white- Gaussian. But in constricted environment, e.g. dense urban environment, because of the signal reflections on the surrounding obstacles, this assumption cannot be used and accuracy and continuity of service of GNSS receivers are strongly degraded. To enhance the localization performances, we propose to use Dirichlet Process Mixtures to model the pseudorange error density at each acquisition step. Next, this estimation will be used in Rao-Blackwellized Particle Filter to compute the position. This sequential estimation is adapted when the noise is non-stationary. This approach will be tested on real data acquired by a single frequency receiver.

On selecting the hyperparameters of the DPM models for the density estimation of observation errors

The Dirichlet Process Mixture (DPM) models represent an attractive approach to modeling latent distributions parametrically. In DPM models the Dirichlet process (DP) is applied especially when the distribution of latent parameters is to be considered as multimodal. DPMs allow for uncertainty in the choice of parametric forms and in the number of mixing components (clusters). The parameters of a DP include the precision a and the base probability measure G0(μ, Σ). In most applications, the choice of priors and posteriors computation for the hyperparameters (α, μ, Σ) clearly influences inferences about the level of clustering in the mixture. This is the main focus of this paper. We consider the problem of density estimation of an observation noise distribution in a dynamic nonlinear model from a Bayesian nonparametric viewpoint. Our approach is illustrated in a real-world data analysis task dealing with the estimation of pseudorange errors in a GNSS based localization context.

Online EM estimation of the Dirichlet process mixtures scale parameter to model the GPS multipath error

The performance of GPS is strongly degraded in a multipath environment. The multipath impact the distribution of the additive noise corrupting the distance measurements between the satellites and the GPS receiver. In this paper, this distribution is assumed unknown and modeled in a flexible way by using the Bayesian non parametric framework and more precisely the Dirichlet process mixtures. Nevertheless, these latter depend on the so-called scale parameter which can be difficult to tune a priori. The originality of our approach consists in adapting a recent version of the online EM algorithm, developed by Cappé for hidden Markov models, to compute a maximum a posteriori estimate of the scale parameter. Then, as the proposed model is non linear and non Gaussian, the EM-based scale parameter estimation is coupled with a Rao-Blackwellized particle filter for the joint estimation of the mobile location and the distance measurement noise distribution.

Enhancement of Galileo and multi-constellation accuracy by modeling pseudorange noises

Satellite-based positioning systems do not offer accurate solutions in urban environments because of propagation disturbances noising measurements. Furthermore, usual positioning computation, like Extended Kalman Filter, relies on considering the noise as a Gaussian Centered distribution, which is unrealistic in constrained area. Moreover, today only the US system (GPS) is fully operational. Available satellite signals are not always sufficient to guarantee a continuity of service in all environments and more particularly in urban environment. However, new satellite navigation systems will appear in a couple of years. These new constellations will provide more satellites and signals. In this paper, at first, we highlight the improvement using GPS in complement with Galileo. Next, we propose a modeling of propagation noises function of the satellite state of reception, and we explain its use in the filtering process. And finally, we show some simulations results.

Advanced signal processing techniques for multipath mitigation in land transportation environment

GNSS (Global Navigation Satellite Systems) are today the core of all new ITS applications and require increasing accuracy in particular in dense urban areas where complex localization scenarios become more and more frequent as the number of users evolves. In order to reduce inaccuracy caused by the obstacles around the receivers antenna, a solution lies in developing new filtering algorithms. These algorithms will take into account both the variation of the noise distribution over time and the reception environment. This paper intends to be a synthesis of our investigations conducted both theoretically and experimentally since 2006. They have been performed by cooperation between Ecole Centrale of Lille and the INRETS-LEOST lab. Results show the achieved performances of each of the proposed algorithms.

On the use of Dirichlet process mixtures for the modelling of pseudorange errors in multi-constellation based localisation

In Global Navigation Satellite Systems (GNSS) positioning, the receiver measures the pseudoranges with respect to each observable navigation satellite and determines the user position. The use of many constellations should lead to highly available, highly accurate navigation anywhere. However, it is important to notice that even if modern receivers achieve high position accuracy in line-of-sight (LOS) conditions, multipath propagation highly degrades positioning performances even in multi-constellation based localisation (”GPS + Galileo” for instance). In urban area, some obstacles (cars, pedestrians, etc) can appear suddenly and thus can induce a random error in the pseudorange measure. We address here the case where the noise probability density functions are of unknown functional form. A flexible Bayesian nonparametric noise model based on Dirichlet process mixtures (DPM) is introduced. Hence, the paper will contain two main parts. The first part focuses on the modelling of the pseudorange noises using DPMs and its suitability in the estimation problem handled by an efficient particle filter. The other part contains interesting validation schemes.