Application Of Blind Source Separation Research Articles

Computing tensor Z-eigenpairs via an alternating direction method.

Tensor eigenproblems have wide applications in blind source separation, magnetic resonance imaging, and molecular conformation. In this study, we explore an alternating direction method for computing the largest or smallest Z-eigenvalue and corresponding eigenvector of an even-order symmetric tensor. The method decomposes a tensor Z-eigenproblem into a series of matrix eigenproblems that can be readily solved using off-the-shelf matrix eigenvalue algorithms. Our numerical results show that, in most cases, the proposed method converges over two times faster and could determine extreme Z-eigenvalues with 20-50% higher probability than a classical power method-based approach.

Online Generalized Eigenvectors Extraction Via a Fixed-Point Approach

Generalized principal component analysis (GPCA) has been an active area of research in statistical signal processing for decades. It is used, e.g., for denoising in subspace tracking as the noise of different nature is incorporated into the procedure of maximizing signal-to-noise ratio (SNR). This paper presents a fixed-point approach concerning the principal generalized eigenvector extraction. It is based on the basis iteration for maximizing the generalized Rayleigh quotient (GRQ) with a given matrix pencil. The proposed approach extracts multiple generalized eigenvectors of a matrix pencil by exploiting the orthogonal complement structure of its estimation. It has no requirement to choose the commonly used step size. This enhances its practical applicability, as selecting an appropriate step size is a bottleneck for most gradient flow based algorithms. Our approach is more suitable for online processing because of its easy implementation and low computational complexity. To show the efficacy, efficiency and practical applicability of the proposed algorithm, we conduct several experiments, two of which concern smart antenna and blind source separation applications. Our simulation results show that the proposed algorithm outperforms several existing algorithms in terms of convergence speed as well as computational time.

Hybrid Source Prior Based Independent Vector Analysis for Blind Separation of Speech Signals

Blind Source Separation (BSS) application is a delinquent issue in a complex reverberant environment with changing room geometric dimensions and an increasing number of speech sources. The BSS application issue is determined by the independent component analysis that usually manipulates higher-order statistical approaches. However, the permutation between desired speech sources remains a challenging issue for BSS applications. The permutation problem is been rectified by Independent Vector Analysis (IVA) for BSS applications in the frequency domain. The performance dependency of the IVA approach solely relies on the selection of appropriate source-prior to preserve the inter-frequency dependencies between the same speech source amongst different frequency bins. Therefore, a hybrid model for the IVA method is presented, which comprises of multivariate generalized Gaussian and super-Gaussian distribution source priors to model low as well as high amplitudes speech signals. The weights of the hybrid model between multivariate Gaussian and generalized Gaussian are assigned in accordance to the energy of the observed non-stationary speech mixture signal. In the simulations, different speech mixtures are generated from various speech sources by simulated room model. The proposed approach evaluates the blind separation performance in terms of signal-to-distortion ratio (SDR) and is compared with well-known BSS methods. The results show an improvement of the proposed methodology for non-stationary speech signals over the state-of-the-art IVA models having a fixed source prior.

Bayesian Nonnegative Matrix Factorization With Dirichlet Process Mixtures

Nonnegative Matrix Factorization (NMF) is valuable in many applications of blind source separation, signal processing and machine learning. A number of algorithms that can infer nonnegative latent factors have been developed, but most of these assume a specific noise kernel. This is insufficient to deal with complex noise in real scenarios. In this paper, we present a hierarchical Dirichlet process nonnegative matrix factorization (DPNMF) model in which the Gaussian mixture model is used to approximate the complex noise distribution. Moreover, the model is cast in the nonparametric Bayesian framework by using Dirichlet process mixture to infer the necessary number of Gaussian components. We derive a mean-field variational inference algorithm for the proposed nonparametric Bayesian model. We first test the model on synthetic data sets contaminated by Gaussian, sparse and mixed noise. We then apply it to extract muscle synergies from the electromyographic (EMG) signal and to select discriminative features for motor imagery single-trial electroencephalogram (EEG) classification. Experimental results demonstrate that DPNMF performs better in extracting the latent nonnegative factors in comparison with state-of-the-art methods.

Single channel high noise level ECG deconvolution using optimized blind adaptive filtering and fixed-point convolution kernel compensation

An electrocardiogram (ECG) is used to record the electrical activity of the heart. However, ECG signals are susceptible to the noise from various sources which increases the probability of misinterpretation and can affect the diagnostic process. Traditional noise cancellation techniques, which uses finite and deterministic coefficient, are not efficient, since the ECG signals are non-stationary. Thus, adaptive filters are commonly utilized on such signal as they can adjust their coefficient according to the changing nature of non-stationary signal. Adaptive algorithms still have a disadvantage that they require the model of noise or desired signal. In this paper, a novel algorithm is introduced based on fixed-point convolution kernel compensation for finding a model for using an adaptive filter; then a recursive least square method is used for completing steps of deconvolution of the ECG signal. The deconvolution method can be used for denoising ECG signals in very low signal to noise ratio circumstances and also can be used in blind source separation applications such as separation of fetal ECG from maternal ECG. ECG signals were utilized in this study are taken from the MIT-BIH Arrhythmia database for showing the performance of the algorithm on denoising applications. The results demonstrate that the proposed algorithm renders a much-improved performance in removing the noise from ECG signals, especially in a scenario where signal to noise ratio is negative. Moreover, the noninvasive fetal ECG dataset (NI-FECG) provided by Physionet is also used for fetal ECG extraction by a single thoracic channel. By comparing fetal ECG extraction methods in the literature and the proposed method, it reveals that the proposed method can extract the QRS complex of fetal ECG by a single thoracic channel as accurate as other methods which use abdominal channels.

Application of blind source separation to the health monitoring of electrical and mechanical faults in a linear actuator

This paper proposes an automated fault isolation and diagnostic chain for the health monitoring of a linear actuator composed of a roller screw driven by a permanent magnet synchronous motor. Four health conditions are considered and diagnosed: the healthy condition, a short circuit in the stator windings, a mechanical backlash in the roller screw, and the combination of both faults. In order to separate the fault signatures, empirical mode decomposition is applied to the motor current, followed by independent component analysis, automatic isolation of the fault signatures, and a classification step for the diagnosis. The novelty proposed consists of an automatic processing of the independent components to isolate the effects of the short-circuit from the effects of the backlash. This isolation step, in contrast to earlier works, requires no human intervention to select signals of interest, making it suitable to real-time onboard diagnostics. Furthermore, results show that independent component analysis occupies an important role in the diagnosis: its omission leads to a reduction in the diagnostic performance of the classifier as well as a reduction in measures of class separability.

A dual fast NLMS adaptive filtering algorithm for blind speech quality enhancement

This paper addresses the problem of acoustic noise reduction and speech enhancement in new telecommunications systems by adaptive filtering algorithms. Recently, a particular attention has been made to the blind source separation (BSS) approach applied for the separation of speech and noise components. The BSS application has inherits the good properties of the adaptive filtering algorithm to give more intelligible enhanced speech signal in term of quality. In this paper, we propose a new dual forward BSS algorithm that is based on signal prediction to give an automatic algorithm with a very fine behavior at the output. This algorithm is called the dual fast normalized least mean square (DFNLMS) algorithm. This algorithm has been tested in various noisy conditions and has shown its superiority in terms of the following objective criteria: cepstral distance (CD), segmental signal-to-noise-ratio (SegSNR), segmental mean square error (SegMSE), and system mismatch (SM). A comparison with other competitive and state-of-the-art algorithms is also presented in this paper.

Blind Source Separation for Unimodal and Multimodal Brain Networks: A Unifying Framework for Subspace Modeling.

In the past decade, numerous advances in the study of the human brain were fostered by successful applications of blind source separation (BSS) methods to a wide range of imaging modalities. The main focus has been on extracting "networks" represented as the underlying latent sources. While the broad success in learning latent representations from multiple datasets has promoted the wide presence of BSS in modern neuroscience, it also introduced a wide variety of objective functions, underlying graphical structures, and parameter constraints for each method. Such diversity, combined with a host of datatype-specific know-how, can cause a sense of disorder and confusion, hampering a practitioner's judgment and impeding further development. We organize the diverse landscape of BSS models by exposing its key features and combining them to establish a novel unifying view of the area. In the process, we unveil important connections among models according to their properties and subspace structures. Consequently, a high-level descriptive structure is exposed, ultimately helping practitioners select the right model for their applications. Equipped with that knowledge, we review the current state of BSS applications to neuroimaging. The gained insight into model connections elicits a broader sense of generalization, highlighting several directions for model development. In light of that, we discuss emerging multi-dataset multidimensional (MDM) models and summarize their benefits for the study of the healthy brain and disease-related changes.

Generic Uniqueness of a Structured Matrix Factorization and Applications in Blind Source Separation

Algebraic geometry, although little explored in signal processing, provides tools that are very convenient for investigating generic properties in a wide range of applications. Generic properties are properties that hold "almost everywhere". We present a set of conditions that are sufficient for demonstrating the generic uniqueness of a certain structured matrix factorization. This set of conditions may be used as a checklist for generic uniqueness in different settings. We discuss two particular applications in detail. We provide a relaxed generic uniqueness condition for joint matrix diagonalization that is relevant for independent component analysis in the underdetermined case. We present generic uniqueness conditions for a recently proposed class of deterministic blind source separation methods that rely on mild source models. For the interested reader we provide some intuition on how the results are connected to their algebraic geometric roots.

Blind Source Separation Applied to Indoor Vehicle Pass-By Measurements

Indoor vehicle pass-by noise applications deal with measuring the exterior noise from a vehicle fixed on a chassis dynamometer in a large hemi-anechoic room. During a standardised acceleration test, the noise is measured with an array of microphones placed in the far-field, and the overall noise level versus vehicle position can be simulated. The indoor facility allows controlled and repeatable measurements independent of weather. For engineering purposes, pass-by contribution analysis can be included in the test leading to information about the pass-by noise contribution from major noise sources. This work presents a novel application of blind source separation to vehicle measurements from an indoor pass-by measurement campaign. In contrast to the classical transfer path approach using point sources for modelling vehicle noise sources and combining an operational measurement with transfer functions, the blind approach does not consider a specific noise source model. It only assumes that the noise is produces by a set of independent noise sources using only a single operational measurement for a given vehicle condition as input. Near-field microphone measurements are blindly separated into independent components and further correlated with the signals measured at the far-field indoor pass-by microphones to get the time-domain contributions. Finally, we apply the indoor simulated pass-by algorithm to produce noise contribution levels as a function of vehicle position. We discuss the specified application of blind source separation to vehicle measurements for different operating conditions from a real indoor pass-by test. Separation of tyre and engine related noise at tyre near-field microphones is verified. Furthermore, the tyre pass-by noise contribution is extracted from the overall vehicle measurement.

Generalized Non-Orthogonal Joint Diagonalization With LU Decomposition and Successive Rotations

Non-orthogonal joint diagonalization (NJD) free of prewhitening has been widely studied in the context of blind source separation (BSS) and array signal processing, etc. However, NJD is used to retrieve the jointly diagonalizable structure for a single set of target matrices which are mostly formulized with a single dataset, and thus is insufficient to handle multiple datasets with inter-set dependences, a scenario often encountered in joint BSS (J-BSS) applications. As such, we present a generalized NJD (GNJD) algorithm to simultaneously perform asymmetric NJD upon multiple sets of target matrices with mutually linked loading matrices, by using LU decomposition and successive rotations, to enable J-BSS over multiple datasets with indication/exploitation of their mutual dependences. Experiments with synthetic and real-world datasets are provided to illustrate the performance of the proposed algorithm.

A high-order statistical tensor based algorithm for anomaly detection in hyperspectral imagery.

Recently, high-order statistics have received more and more interest in the field of hyperspectral anomaly detection. However, most of the existing high-order statistics based anomaly detection methods require stepwise iterations since they are the direct applications of blind source separation. Moreover, these methods usually produce multiple detection maps rather than a single anomaly distribution image. In this study, we exploit the concept of coskewness tensor and propose a new anomaly detection method, which is called COSD (coskewness detector). COSD does not need iteration and can produce single detection map. The experiments based on both simulated and real hyperspectral data sets verify the effectiveness of our algorithm.

Canonical polyadic decomposition of third-order semi-nonnegative semi-symmetric tensors using LU and QR matrix factorizations

Abstract Semi-symmetric three-way arrays are essential tools in blind source separation (BSS) particularly in independent component analysis (ICA). These arrays can be built by resorting to higher order statistics of the data. The canonical polyadic (CP) decomposition of such semi-symmetric three-way arrays allows us to identify the so-called mixing matrix, which contains the information about the intensities of some latent source signals present in the observation channels. In addition, in many applications, such as the magnetic resonance spectroscopy (MRS), the columns of the mixing matrix are viewed as relative concentrations of the spectra of the chemical components. Therefore, the two loading matrices of the three-way array, which are equal to the mixing matrix, are nonnegative. Most existing CP algorithms handle the symmetry and the nonnegativity separately. Up to now, very few of them consider both the semi-nonnegativity and the semi-symmetry structure of the three-way array. Nevertheless, like all the methods based on line search, trust region strategies, and alternating optimization, they appear to be dependent on initialization, requiring in practice a multi-initialization procedure. In order to overcome this drawback, we propose two new methods, called JD LU + and JD QR + , to solve the problem of CP decomposition of semi-nonnegative semi-symmetric three-way arrays. Firstly, we rewrite the constrained optimization problem as an unconstrained one. In fact, the nonnegativity constraint of the two symmetric modes is ensured by means of a square change of variable. Secondly, a Jacobi-like optimization procedure is adopted because of its good convergence property. More precisely, the two new methods use LU and QR matrix factorizations, respectively, which consist in formulating high-dimensional optimization problems into several sequential polynomial and rational subproblems. By using both LU and QR matrix factorizations, we aim at studying the influence of the used matrix factorization. Numerical experiments on simulated arrays emphasize the advantages of the proposed methods especially the one based on LU factorization, in the presence of high-variance model error and of degeneracies such as bottlenecks. A BSS application on MRS data confirms the validity and improvement of the proposed methods.

On the application of blind source separation for damping estimation of bridges under traffic loading

The accurate and reliable estimation of modal damping from output-only vibration measurements of structural systems is a continuing challenge in the fields of operational modal analysis (OMA) and system identification. In this paper a modified version of the blind source separation (BSS)-based Second-Order Blind Identification (SOBI) method was used to perform modal damping identification on a model bridge structure under varying loading conditions. The bridge model was created with finite elements and consisted of a series of stringer beams supported by a larger girder. The excitation was separated into two categories: ambient noise and traffic loads with noise modeled with random forcing vectors and traffic simulated with moving loads for cars and partially distributed moving masses for trains. The acceleration responses were treated as the mixed output signals for the BSS algorithm. The modified SOBI method used a windowing technique to maximize the amount of information used for blind identification from the responses. The modified SOBI method successfully found the mode shapes for both types of excitation with strong accuracy, but power spectral densities (PSDs) of the recovered modal responses showed signs of distortion for the traffic simulations. The distortion had an adverse affect on the damping ratio estimates for some of the modes but no correlation could be found between the accuracy of the damping estimates and the accuracy of the recovered mode shapes. The responses and their PSDs were compared to real-world collected data and patterns similar to distortion were observed implying that this issue likely affects real-world estimates.

Embedded Implementation of Second-Order Blind Identification (SOBI) for Real-Time Applications in Neuroscience

Blind source separation (BSS) is an effective and powerful tool for signal processing and artifact removal in electroencephalographic signals. For real-time applications such as brain–computer interfaces, cognitive neuroscience or clinical neuromonitoring, it is of prime importance that BSS is effectively performed in real time. In order to improve in terms of speed considering the optimal parallelism environment that hardware provides, we build a high-level hardware/software co-simulation based on MATLAB/Simulink for BSS application. To illustrate our approach, we implement the most critical parts of the second-order blind identification algorithm with a fixed-point algorithm on a commercial field-programmable gate array development kit. The results obtained show that co-simulation environment reduces the computation time from 1.9 s to 12.8 ns and thus has great potential for real-time applications.

Research on Application of Blind Source Separation in Rolling Bearing Fault Diagnosis Based on Particle Swarm Optimization

This paper presents blind source separation of rolling bearing based on particle swarm optimization. The algorithm combines the advantages of both blind source separation and particle swarm optimization. Through the experiment it is shown that the algorithm can separate the signals collected from rolling bearing and gearbox effectively, which can provide a new method for fault diagnosis and signal processing of machinery equipment.

Application of Independent Component Analysis in Machine Fault Diagnosis

A novel fast algorithm for lndependent Component Analysis is introduced, which can be used for blind source separation and machine fault diagnosis feature extraction. It is shown how a neural network learning rule can be transformed into a fixed-point iteration, which provides an algorithm that is very simple, does not depend on any user-defined parameters, and is fast to converge to the most accurate solution allowed by the data. The purpose of this paper is to review the application of blind source separation in the machine fault diagnosis,including the following aspects: noise elimination and extraction of the weak signals,the separation of multi-fault sources,redundancy reduction,feature extraction and pattern classification based on independent component analysis. And its application in machine fault diagnosis is illustrated by the examples. In addition, some prospects about using blind source separation for machine fault diagnosis are discussed.

A Review of BSS Based Digital Image Watermarking and Extraction Methods

The field of Signal Processing has witnessed the strong emergence of a new technique, the Blind Signal Processing (BSP) which is based on sound theoretical foundation. An offshoot of the BSP is known as Blind Source Separation (BSS). This digital signal processing techniques have a wide and varied potential applications. The term blind is indicative of the fact that both the source signal and the mixing procedures are unknown. One of the more interesting applications of BSS is in field of image data security/authentication where digital watermarking is proposed. Watermarking is a promising technique to help protect data security and intellectual property rights. The plethora digital image watermarking methods are surveyed and discussed here with their features and limitations. Thus literature survey is presented in two major categories-Digital image watermarking methods and BSS based techniques in digital image watermarking and extraction.

On robustifying some second order blind source separation methods for nonstationary time series

Blind source separation (BSS) is an important analysis tool in various signal processing applications like image, speech or medical signal analysis. The most popular BSS solutions have been developed for independent component analysis (ICA) with identically and independently distributed (iid) observation vectors. In many BSS applications the assumption on iid observations is not realistic, however, as the data are often an observed time series with temporal correlation and even nonstationarity. In this paper, some BSS methods for time series with nonstationary variances are discussed. We also suggest ways to robustify these methods and illustrate their performance in a simulation study.

A “Sequentially Drilled” Joint Congruence (SeDJoCo) Transformation With Applications in Blind Source Separation and Multiuser MIMO Systems

We consider a particular form of the classical approximate joint diagonalization (AJD) problem, which we call a “sequentially drilled” joint congruence (SeDJoCo) transformation. The problem consists of a set of symmetric real-valued (or Hermitian-symmetric complex-valued) target-matrices. The number of matrices in the set equals their dimension, and the joint diagonality criterion requires that in each transformed (“diagonalized”) target-matrix, all off-diagonal elements on one specific row and column (corresponding to the matrix-index in the set) be exactly zeros, yet does not care about the other (diagonal or off-diagonal) elements. The motivation for this form arises in (at least) two different contexts: maximum likelihood blind (or semiblind) source separation and coordinated beamforming for multiple-input multiple-output (MIMO) broadcast channels. We prove that SeDJoCo always has a solution when the target-matrices are positive-definite . We also propose two possible iterative solution algorithms, based on defining and optimizing two different criteria functions, using Newton's method for the first function and successive Jacobi-like transformations for the second. The algorithms' convergence behavior and the attainable performance in the two contexts above are demonstrated in simulation experiments.