Blind Signal Separation Method Research Articles

A two-stage dual-matrix method of blind signal separation

In this paper, we propose a dual-matrix method (DMM) that uses constrained terms to reflect the numerical relation between a mixing matrix and a separating matrix. Thus, some undesired solutions are excluded, the search region is reduced, and the convergent efficiency of the algorithm is ultimately improved. Moreover, DMM is proven to converge in a separable matrix only if its cost function converges to zero. Computer simulations indicate that the algorithm excellently performs blind signal separation; moreover, theoretical analysis matches the simulated results. On the other hand, two-separating system is proposed in the separating stage, and a new performance index (PI) is presented as well. Further, the theoretical properties of the new PI are proven, and the simulations indicate that the new PI is more practical than the commonly used one. Two numerical simulations were performed in the current study to illustrate the efficiency of the proposed method.

Direction-of-Arrival Estimation and Sensor Array Error Calibration Based on Blind Signal Separation

We consider estimating the direction-of-arrival (DOA) in the presence of sensor array error. In the proposed method, a blind signal separation method, the joint approximation and diagonalization of eigenmatrices algorithm, is implemented to separate the signal vector and the mixing matrix consisting of the array manifold matrix and the sensor array error matrix. Based on a new mixing matrix and the reconstruction of the array output vector of each individual signal, we propose a novel DOA estimation and sensor array error calibration procedure. This method is independent of array phase errors and performs well against difference of SNR of signals. Numerical simulations verify the effectiveness of the proposed method.

Limitation of the Lateral Angled Broadband Low Frequency Impact Excitation on the Non-Destructive Condition Assessment of the Timber Utility Poles

Timber utility poles play a significant role in the infrastructure of Australia as well as many other countries for power distribution and communication networks. Due to the advanced age of Australia’s timber pole infrastructure, substantial efforts are undertaken on maintenance and asset management to avoid any failures of the utility lines. Nevertheless, the lack of reliable tools for assessing the condition of in-service poles seriously jeopardizes the maintenance and asset management. For instance, each year approximately 300,000 poles are replaced in the Eastern States of Australia with up to 80% of them still being in a very good condition, resulting in major waste of natural resources and money. Non-destructive testing (NDT) methods based on stress wave propagation can potentially offer simple and cost-effective tools for identifying the in-service condition of timber poles. Nonetheless, most of the currently available methods are not appropriate for condition assessment of timber poles in-service due to presence of uncertainties such as complicated material properties, environmental conditions, interaction of soil and structure, and an impact excitation type. In order to address these complexities, advanced digital signal processing methodologies are needed to be employed. Deterministic signal separation, blind signal separation, and frequency-wavenumber velocity filtering are the three groups of methodologies, which could most probably provide solutions. In this paper applicability and effectiveness of the blind signal separation methods is investigated through a numerical data obtained from of a timber pole modelled with both isotropic and orthotropic material properties. Principal Component Analysis (PCA), Singular Value Decomposition (SVD), and K-means clustering algorithms are the blind signal separation methodologies that are employed in this research work.

A novel method for blind signal separation of single-channel and time-frequency overlapped multi-component signal

In this paper, a novel blind signal separation BSS method of single-input multi-output SIMO system is proposed based on the periodicity of independent components, which can separate time-frequency overlapped multi-component signal effectively. It adopts the technique of changing single-channel recordings vector into virtual multi-channel measurement matrix. Singular value decomposition SVD method is utilised to decompose the matrix. Singular value ratio SVR spectrum can be calculated by changing the virtual multi-channel measurement matrix. Combining SVR spectrum and the periodic characteristic of independent components, the independent components are isolated successfully. The source number can be estimated by the peak value of SVR spectrum. The simulation results demonstrate the effectiveness of the proposed algorithm in this paper and show the method can guarantee high separation accuracy in low signal-noise-ratio SNR environment.

Remote diffuse reflectance spectroscopy sensor for tissue engineering monitoring based on blind signal separation.

In this study the first results on evaluation and assessment of grafted bioengineered skin substitutes using an optical Diffuse Reflectance Spectroscopy (DRS) system with a remote optical probe are shown. The proposed system is able to detect early vascularization of skin substitutes expressing the Vascular Endothelial Growth Factor (VEGF) protein compared to normal grafts, even though devitalized skin is used to protect the grafts. Given the particularities of the biological problem, data analysis is performed using two Blind Signal Separation (BSS) methods: Principal Component Analysis (PCA) and Independent Component Analysis (ICA). These preliminary results are the first step towards point-of-care diagnostics for skin implants early assessment.

Gradient method for blind chaotic signal separation based on proliferation exponent

A new method to perform blind separation of chaotic signals is articulated in this paper, which takes advantage of the underlying features in the phase space for identifying various chaotic sources. Without incorporating any prior information about the source equations, the proposed algorithm can not only separate the mixed signals in just a few iterations, but also outperforms the fast independent component analysis (FastICA) method when noise contamination is considerable.

Kernel Function and Parameters Optimization in KICA for Rolling Bearing Fault Diagnosis

Kernel independent component analysis (KICA) is a blind signal separation method which has a good effect for the treatment of non-linear signal. For introducing kernel techniques, the choices of kernel function and its kernel parameter have a great influence on the analytic results. A kernel function and its parameters optimization method is proposed on the basis of the similarity of source fault signals and kernel independent component. The similarity parameter is proposed to verify the merits or defects of KICA by using different kernel function and parameters. The simulation studies are processed, and the simulation conclusion is verified by the actual diagnostic case. These provide guidance for the application of the KICA method in the mechanical fault diagnosis.

SU‐E‐J‐25: A Quantitative Comparison of Contrast‐To‐Noise in Rib Suppressed Images

Purpose: To compare the contrast to noise ratio (CNR) of two different rib suppression methodologies ‐ dual‐energy (DE) image subtraction and blind signal separation (BSS)‐used for lung cancer patients receiving image‐guided radiotherapy. Methods: A database consisting of 28 planar radiographs obtained at 60 kVp and 120 kVp were used in this study. DE subtraction imaging was performed by taking a weighted logarithmic subtraction of the individual images. A weighting parameter of 2.1 was chosen to produce the optimal soft tissue (rib suppressed) images for each image pair. Similarly, BSS was used to extract the soft tissue and bony image components from a patient image. An open source program for performing BSS, called FastICA, was used in conjunction with the same 60 kVp and 120 kVp images. Regions of interest (ROIs) were contoured on the resultant images and the CNR from each technique was calculated and compared. Results: A paired t‐test was run on the CNR values, resulting in a p value < 0.05, indicated there was a statistically significant difference in the CNRs generated by the two methods. For 24 of the 28 image sets that were evaluated, the CNR values were found to be higher for the DE subtraction images. Of these 24 images, 9 had CNR values > 20% higher, and 4 had CNR values > 40% higher. Conclusion: Based on this study, DE image subtraction creates images with higher CNR values than those generated with FastICA. This is likely due, at least in part, to the ability to customize the weighting factor of the DE subtraction images. However, there are numerous methods for performing BSS outside of FastICA, and further study is warranted to determine if alternative BSS methods could produce better rib suppressed images. This research was supported by Varian Medical Systems.

Geometric method for blind signal separation of multichannel MPSK signals

In this letter, the problem of blind source separation of Multiple-Phase-Shift-Keying (MPSK) digital signal is considered. The geometry of received MPSK signals constellation is exploited. The column vectors of received signals can be regarded as the points of hyper-cube. All the possible distinct vectors of received signals are found by clustering, and mixing matrix and sources are estimated by searching out the pairing vectors and eliminating redundant information in all possible distinct vectors. Simulation results give the polar diagram of estimated original signals. They show that the proposed algorithm is effective when the original signals is Quadrature-Phase-Shift-Keying (QPSK) or 8-Phase-Shift-Keying (8PSK).

Coupled blind signal separation and spectroscopic database fitting of the mid infrared PAH features

The aromatic infrared bands (AIBs) observed in the mid infrared spectrum are attributed to Polycyclic Aromatic Hydrocarbons (PAHs). We observe the NGC 7023-North West (NW) PDR in the mid-infrared (10 - 19.5 micron) using the Infrared Spectrometer (IRS), on board Spitzer. Clear variations are observed in the spectra, most notably the ratio of the 11.0 to 11.2 micron bands, the peak position of the 11.2 and 12.0 micron bands, and the degree of asymmetry of the 11.2 micron band. The observed variations appear to change as a function of position within the PDR. We aim to explain these variations by a change in the abundances of the emitting components of the PDR. A Blind Signal Separation (BSS) method, i.e. a Non-Negative Matrix Factorization algorithm is applied to separate the observed spectrum into components. Using the NASA Ames PAH IR Spectroscopic Database, these extracted signals are fit. The observed signals alone were also fit using the database and these components are compared to the BSS components. Three component signals were extracted from the observation using BSS. We attribute the three signals to ionized PAHs, neutral PAHs, and Very Small Grains (VSGs). The fit of the BSS extracted spectra with the PAH database further confirms the attribution to ionized and neutral PAHs and provides confidence in both methods for producing reliable results. The 11.0 micron feature is attributed to PAH cations while the 11.2 micron band is attributed to neutral PAHs. The VSG signal shows a characteristically asymmetric broad feature at 11.3 micron with an extended red wing. By combining the NASA Ames PAH IR Spectroscopic Database fit with the BSS method, the independent results of each method can be confirmed and some limitations of each method are overcome.

Source localization for multiple speech sources using low complexity non-parametric source separation and clustering

This article presents a new method for localization of multiple concurrent speech sources that relies on simultaneous blind signal separation and direction of arrival (DOA) estimation, as well as a method to solve the intersection point selection problem that arises when locating multiple speech sources using multiple sensor arrays. The proposed method is based on a low complexity non-parametric blind signal separation method, making is suitable for real-time applications on embedded platforms. On top of reduced complexity in comparison to a previously presented method, the DOA estimation accuracy is also improved. Evaluation of the performance is done with both real recording and simulations, and a real-time prototype of the proposed method has been implemented on a DSP platform to evaluate the computational and the memory complexities in a real application.

Blind Extraction of Global Signal From Multi-Channel Noisy Observations

We propose a novel efficient method of blind signal extraction from multi-sensor networks when each observed signal consists of one global signal and local uncorrelated signals. Most of existing blind signal separation and extraction methods such as independent component analysis have constraints such as statistical independence, non-Gaussianity, and underdetermination, and they are not suitable for global signal extraction problem from noisy observations. We developed an estimation algorithm based on alternating iteration and the smart weighted averaging. The proposed method does not have strong assumptions such as independence or non-Gaussianity. Experimental results using a musical signal and a real electroencephalogram demonstrate the advantage of the proposed method.

Independent Subspace Analysis For Blind Signal Separation: Models And Algorithms

As an extended independent component analysis (ICA) method for blind signal separation (BSS), independent subspace analysis (ISA) has more applications than ICA method. In this paper, we briefly present a new perspective of ISA for BSS. The general and detailed definition of the ISA model is given, the relationships between ICA and ISA methods is also discussed. Moreover, due to the fundamental difficulty in the ISA problem that it (specify it here) is not unique without extra constraints, we also review and discuss the separateness and uniqueness of the ISA model. At last, the state-of-art ISA algorithms are overviewed from different theoretical foundations. Some ISA algorithms based on the original relative gradient (natural gradient) ICA, FastICA, and JADE ICA are constructed for the BSS problem in detail and simulations of these algorithms are also exhibited.

INDEPENDENT SUBSPACE ANALYSIS FOR BLIND SIGNAL SEPARATION: MODELS AND ALGORITHMS

AbstractAs an extended independent component analysis (ICA) method for blind signal separation (BSS), independent subspace analysis (ISA) has more applications than ICA method. In this paper, we briefly present a new perspective of ISA for BSS. The general and detailed definition of the ISA model is given, the relationships between ICA and ISA methods is also discussed. Moreover, due to the fundamental difficulty in the ISA problem that it (specify it here) is not unique without extra constraints, we also review and discuss the separateness and uniqueness of the ISA model. At last, the state-of-art ISA algorithms are overviewed from different theoretical foundations. Some ISA algorithms based on the original relative gradient (natural gradient) ICA, FastICA, and JADE ICA are constructed for the BSS problem in detail and simulations of these algorithms are also exhibited.

Blind signal separation of linear mixture using trilinear decomposition

This paper introduces a new source separation technique exploiting the time coherence of the source signals. The proposed approach relies only on stationary second order statistics. Blind Signal Separation (BSS) method using trilinear decomposition is proposed in this paper. Simulation results reveal that our proposed algorithm has the better blind signal separation performance than joint diagonalization method. Our proposed algorithm does not require whitening processing. Moreover, our proposed algorithm works well in the underdetermined condition, where the number of sources exceeds than the number of sensors.

Convolutive blind signal separation by estimating mixing channels in time domain

Unlike conventional blind signal separation (BSS) methods estimating unmixing channels, an improved BSS method by estimating mixing channels in time domain is presented. Compared to the unmixing channels with IIR characteristics the mixing channels can be approximated by FIR filters with fewer time delays, and the proposed method demonstrates superior performance for the same training data.

Searching-and-averaging method of underdetermined blind speech signal separation in time domain

Underdetermined blind signal separation (BSS) (with fewer observed mixtures than sources) is discussed. A novel searching-and-averaging method in time domain (SAMTD) is proposed. It can solve a kind of problems that are very hard to solve by using sparse representation in frequency domain. Bypassing the disadvantages of traditional clustering (e.g., K-means or potential-function clustering), the durative- sparsity of a speech signal in time domain is used. To recover the mixing matrix, our method deletes those samples, which are not in the same or inverse direction of the basis vectors. To recover the sources, an improved geometric approach to overcomplete ICA (Independent Component Analysis) is presented. Several speech signal experiments demonstrate the good performance of the proposed method.

Reply to “Comments on `A Blind Signal Separation Method for Multiuser Communications'”

In this correspondence, we reply to the comments in Gu ["Comments on `A blind signal separation method for multiuser communications'", IEEE Transactions on Signal Processing, vol. 55, no. 5, pp. 2355-2356, May 2007]. We explain that the global convergence analysis of the method in Castedo ["A blind signal separation method for multiuser communications", IEEE Transactions on Signal Processing, vol. 45, no. 5, pp. 1343-1348, May 1997] can be completed as an extension of the results that were already presented. Also, we recall that two of the authors put forward (Dapena and Castedo ["Stochastic gradient adaptive algorithms for blind source separation", Elsevier Signal Processing, vol. 75, pp. 11-27]) a similar method with more attractive properties whose convergence analysis is not affected by the counterexample provided in Gu et al

Analysis of the emission of very small dust particles from Spitzer spectro-imagery data using blind signal separation methods

This work was conducted as part of the SPECPDR program, dedicated to the study of very small particles and astrochemistry, in Photo-Dissociation Regions (PDRs). We present the analysis of the mid-IR spectro-imagery observations of Ced 201, NCG 7023 East and North-West and rho-Ophiuchi West filament. Using the data from all four modules of the InfraRed Spectrograph onboard the Spitzer Space Telescope, we produced a spectral cube ranging from 5 to 35 um for each one of the observed PDRs. The resulting cubes were analysed using Blind Signal Separation methods (NMF and FastICA). For Ced 201, rho-Ophiuchi West filament and NGC 7023 East, we find that two signals can be extracted from the original data cubes, which are 5 to 35 um spectra. The main features of the first spectrum are a strong continuum emission at long wavelengths, and a broad 7.8 um band. On the contrary, the second spectrum exhibits the classical Aromatic Infrared Bands (AIBs) and no continuum. The reconstructed spatial distribution maps show that the latter spectrum is mainly present at the cloud surface, close to the star whereas the first one is located slightly deeper inside the PDR. The study of the spectral energy distribution of Ced 201 up to 100 um suggests that, in cool PDRs, the 5-25 um continuum is carried by Very Small Grains (VSGs). The AIB spectra in the observed objects can be interpreted as the contribution of neutral and positively-charged Polycyclic Aromatic Hydrocarbons (PAHs). We extracted the 5 to 25 um emission spectrum of VSGs in cool PDRs, these grains being most likely carbonaceous. We show that the variations of the mid-IR (5-35 um) spectra of PDRs can be explained by the photo-chemical processing of these VSGs and PAHs, VSGs being the progenitors of free PAH.

A method for blind separation of components information from mixed pixel*

Abstract In the field of remote sensing, it is important to separate the component information from mixed pixel. If the physical process of remote sensing can be expressed by a set of linear equations, the remote sensing information matrix is equal to the weight matrix multiplied by the component information matrix. Generally speaking, the precondition of retrieval of component information matrix is that the weight matrix is known. However, the blind signal separation (BSS) method can separate the matrix unconditionally, whose basic principle is that the additive information needing separation can be achieved from the statistical characteristics contained in a mass of samples in the remotely sensed information matrix. Therefore, the values of the component information matrix and the weight matrix can be estimated. The wave shape f components can be retrieved by BSS, but the amplitude cannot. In this paper, the plant-soil mixed pixels were chosen as the studying targets in this paper t quantitatively separ...