Robust Estimation of Directions-of-Arrival in Diffuse Noise Based on Matrix-Space Sparsity
نویسندگان
چکیده
We consider the estimation of the Directions-Of-Arrival (DOA) of target signals in diffuse noise. The state-of-the-art MUltiple SIgnal Classification (MUSIC) algorithm necessitates accurate identification of the signal subspace. In diffuse noise, however, it is difficult to identify it directly from the observed spatial covariance matrix. In our approach, we estimate the target spatial covariance matrix, so that we can identify the orthogonal complement of the signal subspace as its null space. We present a unified framework for modeling noise covariance in a matrix space, which generalizes four state-of-the-art diffuse noise models. We propose two alternative algorithms for estimating the target spatial covariance matrix, namely Low-rank Matrix Completion (LMC) and Trace Norm Minimization (TNM). These rely on denoising of the observed spatial covariance matrix via orthogonal projection onto the orthogonal complement of the noise matrix subspace. The missing component lying in the noise matrix subspace is then completed by exploiting the low-rankness of the target spatial covariance matrix. Large-scale experiments with real-world noise show that TNM with a certain noise model outperforms conventional MUSIC based on Generalized EigenValue Decomposition (GEVD) by 5% in terms of the precision averaged over the dataset. Key-words: diffuse noise, DOA estimation, microphone arrays, MUSIC, matrix completion. ∗ Nobutaka Ito is with NTT Communication Science Labs. He performed this work as a joint PhD student with the University of Tokyo and University of Rennes 1. † Nobutaka Ono is with the National Institute of Informatics, Japan. ‡ Shigeki Sagayama is with the Universito of Tokyo. ha l-0 07 46 27 1, v er si on 1 28 O ct 2 01 2 Estimation Robuste de Directions d’Arrivée dans du Bruit Diffus Basée sur la Parcimonie dans un Espace Matriciel Résumé : Nous considérons l’estimation des directions d’arrivée de sources sonores dans du bruit diffus. L’algorithme de l’état de l’art MUSIC (MUltiple SIgnal Classification) nécessite l’identification précise du sous-espace signal. En présence de bruit diffus, cependant, il est difficile de l’estimer directement à partir de la matrice de covariance spatiale observée. Dans notre approche, nous estimons la matrice de covariance spatiale de la source cible, de sorte à pouvoir identifier le complément orthogonal du sous-espace signal comme son espace nul. Nous présentons un cadre unifié pour la modélisation de la matrice de covariance du bruit dans un espace matriciel, qui généralise quatre modèles de bruit diffus de l’état de l’art. Nous proposons deux algorithmes pour estimer la matrice de covariance spatial de la cible, basés soit sur la complétion de matrice de rang faible soit sur la minimisation de la norme trace. Ces algorithmes reposent sur le débruitage de la matrice de covariance spatiale observée par projection orthogonale sur le complément du sous-espace matriciel correspondant au bruit. La composante manquante dans le sous-espace matriciel correspondant au bruit est alors complétée en utilisant le faible rang de la matrice de covariance spatiale de la cible. Des expériences à grande échelle montrent que, pour l’un des modèles de bruit, la minimisation de la norme trace dépasse l’approche classique par MUSIC avec décomposition en valeurs propres généralisée de 5% en terme de précision en moyenne. Mots-clés : bruit diffus, estimation de direction d’arrivée, antenne de microphones, MUSIC, complétion de matrice. ha l-0 07 46 27 1, v er si on 1 28 O ct 2 01 2 Robust Estimation of Directions-of-Arrival in Diffuse Noise 3
منابع مشابه
جهت یابی چند گوینده با استفاده از روش WCSSDOA
In this paper we propose the spatial sparsity based WCSSDOA method for multi speakers' Direction of arrival estimation. In the proposed method the sparse modeling is done based on the sensor signals' correlation matrix, which leads to low computational complexity. In this method the SVD decomposition of the noise covariance matrix is proposed to reach the free noise sparse model, which leads to...
متن کاملCrystal-MUSIC: Accurate Localization of Multiple Sources in Diffuse Noise Environments Using Crystal-Shaped Microphone Arrays
This paper presents crystal-MUSIC, a method for DOA estimation of multiple sources in the presence of diffuse noise. MUSIC is well known as a method for the estimation of the DOAs of multiple sources but is not very robust to diffuse noise from many directions, because the covariance structure of such noise is not spherical. Our method makes it possible for MUSIC to accurately estimate the DOAs...
متن کاملSpeech Enhancement using Adaptive Data-Based Dictionary Learning
In this paper, a speech enhancement method based on sparse representation of data frames has been presented. Speech enhancement is one of the most applicable areas in different signal processing fields. The objective of a speech enhancement system is improvement of either intelligibility or quality of the speech signals. This process is carried out using the speech signal processing techniques ...
متن کاملRobust Attitude Control of Spacecraft Simulator with External Disturbances
The spacecraft simulator robust control through H∞-based linear matrix inequality (LMI) and robust adaptive method is implemented. The spacecraft attitude control subsystem simulator consists of a platform, an air-bearing and a set of four reaction wheels. This set up provides a free real-time three degree of freedom rotation. Spacecraft simulators are applied in upgrading and checking the c...
متن کاملRobust Identification of Smart Foam Using Set Mem-bership Estimation in A Model Error Modeling Frame-work
The aim of this paper is robust identification of smart foam, as an electroacoustic transducer, considering unmodeled dynamics due to nonlinearities in behaviour at low frequencies and measurement noise at high frequencies as existent uncertainties. Set membership estimation combined with model error modelling technique is used where the approach is based on worst case scenario with unknown but...
متن کامل