Showing papers by "Karim Abed-Meraim published in 2001"
01 Jan 2001
TL;DR: A new separation method based on time-frequency distributions (TFDs) for blind separation of nonstationary sources in the underdetermined case, where there are more sources than sensors is proposed.
Abstract: This paper deals with the problem of blind source separation of nonstationary signals of which only instantaneous linear signals are observed. Exploiting the effectiveness of time-frequency signal processing for nonstationary signals, a blind source separation approach is considered using the observation spatial time-frequency distributions (STFD). Existing solutions are bound to the situation in which the number of sources being separated is less than the number of available sensors measuring the mixed sources. We consider the more general case when we can have more sources than sensors assuming that the former are "separable" in the time-frequency domain. The proposed solution proceeds through 3 main steps: (i) a testing procedure is applied (after whitening the STFD) to first separate the cross-terms from auto-terms; (ii), source separation in the time-frequency domain (from the autoterms only) using a vector classification approach; and finally (iii), obtaining the source signatures using time-frequency synthesis.
182 citations
TL;DR: This paper studies the blind source separation (BSS) problem with the assumption that the source signals are cyclostationary, and an iterative algorithm is derived to minimize this contrast function.
Abstract: This paper studies the blind source separation (BSS) problem with the assumption that the source signals are cyclostationary. Identifiability and separability criteria based on second-order cyclostationary statistics (SOCS) alone are derived. The identifiability condition is used to define an appropriate contrast function. An iterative algorithm (ATH2) is derived to minimize this contrast function. This algorithm separates the sources even when they do not have distinct cycle frequencies.
125 citations
TL;DR: Two normalized versions of Oja's (1992) algorithm (NOja and NOOja), which can be used for the estimation of minor and principal subspaces of a vector sequence, offer a faster convergence, orthogonality, and a better numerical stability.
Abstract: We present two normalized versions of Oja's (1992) algorithm (NOja and NOOja), which can be used for the estimation of minor and principal subspaces of a vector sequence. The new algorithms offer, as compared to Oja, a faster convergence, orthogonality, and a better numerical stability with a slight increase in computational complexity.
84 citations
07 May 2001
TL;DR: A blind source separation approach exploiting both auto-terms and cross-terms of the time-frequency (TF) distributions of the sources is considered, based on the simultaneous diagonalization and anti-diagonalization of spatial TF distribution matrices made up of auto- terms andCross-terms.
Abstract: We address the problem of blind source separation of non-stationary signals of which only instantaneous linear mixtures are observed. A blind source separation approach exploiting both auto-terms and cross-terms of the time-frequency (TF) distributions of the sources is considered. The approach is based on the simultaneous diagonalization and anti-diagonalization of spatial TF distribution matrices made up of, respectively, auto-terms and cross-terms. Numerical simulations are provided to demonstrate the effectiveness of the proposed approach and compare its performances with existing TF-based methods.
71 citations
24 Jun 2001
TL;DR: The sphere decoder algorithm offers a very powerful tool to reach near the maximum likelihood (ML) decoding performance in several cases such as lattice codes decoding over the Gaussian and Rayleigh fading channels, multiuser detection, uncoded multi-antenna systems detection and space-time codes decoding, and vector quantization.
Abstract: The fast development of digital communications hardware allows for the application of very powerful algorithms at the expense of a small increase in complexity compared to the traditionally implemented algorithms. In this paper we give further results on the sphere decoder (SD) algorithm, and its applications to a broad range of digital communications problems related to the separation of m independent sources by n sensors. First, we discuss practical implementation issues and propose an efficient method to initialize the SD parameters based on computing an estimate of the packing radius of the lattice. We relate the initializing method to the expected performance of the SD, and show that at high SNR, one obtains near optimum performance. The complexity of the SD is then shown to be much less than the upper bound on the complexity of the Fincke and Pohst (1985) algorithm for the problem of finding short length vectors in an m-dimensional lattice. Simulations show that the SD of an m-dimensional lattice needs at most O(m/sup 4.5/) arithmetic operations at low SNR, and O(m/sup 3/) at high SNR. The obtained results offer a very powerful tool to reach near the maximum likelihood (ML) decoding performance in several cases such as lattice codes decoding over the Gaussian and Rayleigh fading channels, multiuser detection, uncoded multi-antenna systems detection and space-time codes decoding, and vector quantization.
68 citations
TL;DR: An iterative detection algorithm of an uncoded multi-transmitter multi-receiver system that is 4 to 8 times less complex than that of the V-BLAST OPT, while maintaining comparable performance.
Abstract: We study an iterative detection algorithm of an uncoded multi-transmitter multi-receiver system. The main data stream is demultiplexed into M substreams, and each substream is modulated independently then transmitted by its dedicated antenna. The receiver disposes of N ≥ M antennas. Over each receive antenna the signal is a superposition of the M substreams affected by independent fades and disturbed by AWGN. The detection algorithm is based on the QR decomposition of the channel transfer matrix which is then used to perform hard or soft inter-substream interference cancellation. Comparisons are done with the V-BLAST optimal order (OPT) detection algorithm. The proposed algorithm is 4 to 8 times less complex than that of the V-BLAST OPT, while maintaining comparable performance.
56 citations
TL;DR: In this article, a robust solution referred to as second order blind identification (SOBI) has been proposed for the case of convolutive mixtures, and this technique is extended to the convolutive mixture case.
Abstract: A new solution to the blind recovery of independent source signals from their convolutive mixtures is presented. In the case of instantaneous mixtures, a robust solution referred to as second order blind identification (SOBI) has been proposed previously; here, this technique is extended to the convolutive mixture case.
44 citations
TL;DR: It is shown that the parametric subspace method gives consistent channel estimates when only an upper bound of the channel order is known, and the proposed algorithm is second-order optimal for a large class of channels.
Abstract: In this paper, blind identification of single-input multiple-output (SIMO) systems using second-order statistics (SOS) only is considered. Using the assumption of a specular multipath channel, we investigate a parametric variant of the so-called subspace method. Nonparametric subspace-based methods require precise estimation of the model order; overestimation of the model order leads to inconsistent channel estimates. We show that the parametric subspace method gives consistent channel estimates when only an upper bound of the channel order is known. A new algorithm, which exploits parametric information on the channel structure, is presented. A statistical performance analysis of the proposed parametric subspace criterion is presented; limited Monte Carlo experiments show that the proposed algorithm is second-order optimal for a large class of channels.
37 citations
Patent•
07 Feb 2001TL;DR: In this article, a Gaussian probability density function (GPMF) is used to transform a multichannel modulation (MCM) signal to a uniform GPDF, which results in a reduced peak-to-average power ratio (PAPR).
Abstract: A probability distribution transformer (110) in a multi-carrier modulation (MCM) transmitter (101) receives a MCM signal comprising data packets that represent amplitude values, where the amplitude values are characterized by a Gaussian probability density function The probability distribution transformer (110), which is provided by a number of piecewise linear transforms, produce a transformed MCM signal comprising transformed data packets which represent transformed amplitude values, where the transformed amplitude values are characterized by a uniform probability density function When transmitted, the transformed MCM signal results in reduced peak-to-average power ratio (PAPR) In a corresponding MCM receiver (102), a probability distribution inverter (180) inverts the transformation
33 citations
06 Aug 2001
TL;DR: In this article, a joint block-diagonalization of positive spatio-temporal covariance matrices is proposed to recover independent source signals from their convolutive mixtures without any a priori knowledge on their structure.
Abstract: Recovering independent source signals from their convolutive mixtures without any a priori knowledge on their structure represents a great challenge in signal processing. We present an efficient solution that is based on the joint block-diagonalization of positive spatio-temporal covariance matrices. In the case of instantaneous mixtures, robust solutions have been proposed previously. Taking advantage of possible non-stationarity of the sources, this new technique uses only second order statistics. The new approach has been successfully applied to the separation of speech signals.
25 citations
01 Jan 2001
TL;DR: The proposed method has been successfully applied to the separation of speech signals and an algorithm that performs a joint block diagonalization of spatio-temporal correlation matrices instead of joint diagonalization as in SOBI is proposed.
Abstract: We present an efficient and new solution to the recovery of independent source signals from their convolutive mixtures without any a priori knowledge on the mixtures. In the case of instantaneous mixtures, a robust solution referred to as second order blind identification (SOBI) has been proposed previously. We extend this technique to the convolutive mixture case and we propose an algorithm that performs a joint block diagonalization of spatio-temporal correlation matrices instead of joint diagonalization as in SOBI. The proposed method has been successfully applied to the separation of speech signals.
06 Aug 2001
TL;DR: This work aims to estimate closely spaced DOAs in array processing using the two step-MUSIC method, a joint estimation strategy (JES) similar to that proposed by Gershman et al. (1996).
Abstract: High resolution methods such as MUSIC fail to separate closely spaced sources in difficult contexts (low SNR, short sample size,...). Halder et al.(1997) have applied an interleaving technique to improve the resolution as well as the performances in the case of frequency estimation. Here we extend this work and deal with the application of this technique to array processing. We aim to estimate closely spaced DOAs. After a first estimation with MUSIC, a second step of the algorithm consists in refining the angle resolution using downsampled covariance matrices together with a joint estimation strategy (JES) similar to that proposed by Gershman et al. (1996). This method improves MUSIC performances especially for low SNRs. Simulations examples are provided to illustrate the performance of the proposed method referred to as two step-MUSIC (TS-MUSIC).
07 May 2001
TL;DR: A new signal processing method is developed for solving the multiline fitting problem in a two dimensional image that is able to estimate the parameters of parallel lines with different offsets and handles the quantization noise effect.
Abstract: A new signal processing method is developed for solving the multiline fitting problem in a two dimensional image. We first reformulate the former problem in a special parameter estimation framework such that a first order or a second order polynomial phase signal structure is obtained. Then, the previously developed algorithms in that formalism (and particularly the downsampling technique for high resolution frequency estimation) can be exploited to produce accurate estimates for line parameters. This method is able to estimate the parameters of parallel lines with different offsets and handles the quantization noise effect which can not be done by the sensor array processing technique introduced by Aghajan et al. (1993). Simulation results are presented to demonstrate the usefulness of the proposed method.
20 Mar 2001
TL;DR: Two normalized versions of the Oja (1992) algorithm are presented which can be used for the estimation of minor and principal subspaces of a vector sequence and offer a faster convergence, better orthogonality and numerical stability with a slight increase in computational complexity.
Abstract: We present two normalized versions of the Oja (1992) algorithm (NOja and NOOja) which can be used for the estimation of minor (noise) and principal (signal) subspaces of a vector sequence. The new algorithms offer, as compared to Oja, a faster convergence, a better orthogonality and numerical stability with a slight increase in computational complexity. These algorithms can find many applications, in particular, in wireless communications.
06 Aug 2001
TL;DR: The orthogonal minimum subspace (OMNS) method is more efficient in computation than a standard subspace method, and is more robust to channel noise than MNS.
Abstract: This contribution deals with a particular family of blind system identification techniques, referred to as minimum noise subspace (MNS) method. The MNS method is a computationally fast version of the subspace method. We develop an orthogonal version of MNS method. The orthogonal minimum subspace (OMNS) method is more efficient in computation than a standard subspace method, and is more robust to channel noise than MNS.
01 Jan 2001
TL;DR: Results obtained by simulation show that the proposed detector significantly improves the performance of other multi-user detectors and even more it improves the single user bound for Gaussian noise at the cost of a slight increase in the implementation complexity.
Abstract: Experimental measurements have demonstrated that in real systems many physical communication channels are non-Gaussian. In this paper, we propose a novel multi-stage multi-user detector that mitigates the MAI and rejects the effects of the non-Gaussian noise in a DS-CDMA system. Results obtained by simulation show that the proposed detector significantly improves the performance of other multi-user detectors and even more it improves the single user bound for Gaussian noise at the cost of a slight increase in the implementation complexity.
01 Jan 2001
TL;DR: The proposed detector can be viewed as an efficient adaptive implementation of the batch-processing subspace-based method proposed by Torlak and Xu and the theoretical and experimental justification of the robustness of the method against channel order overestimation errors are given.
Abstract: This work deals with the problem of signal reception in code division multiple access (CDMA) systems. We present an adaptive detector for multipath CDMA channels which mitigates the effects of multipath propagation (interchip and intersymbol interference) and multiple-access interference. It can be suitably implemented in mobile terminals of the CDMA systems due to its low complexity and single-user structure. The proposed detector can be viewed as an efficient adaptive implementation of the batch-processing subspace-based method proposed by Torlak and Xu (see IEEE Trans. on SP, vol.45, no.1, p.137-47, 1997). In this work, we also give the theoretical and experimental justification of the robustness of the method against channel order overestimation errors.
01 Jan 2001
TL;DR: It is shown that the parametric subspace method gives consistent channel estimates when only an upper bound of the channel order is known, and is second-order optimal for a large class of channels.
Abstract: In this paper, blind identification of single-input mul- tiple-output (SIMO) systems using second-order statistics (SOS) only is considered. Using the assumption of a specular multipath channel, we investigate a parametric variant of the so-called sub- space method. Nonparametric subspace-based methods require a precise estimation of the model order; overestimation of the model order leads to inconsistent channel estimates. We show that the parametric subspace method gives consistent channel estimates when only an upper bound of the channel order is known. A new algorithm, which exploits parametric information on the channel structure, is presented. A statistical performance analysis of the proposed parametric subspace criterion is presented; limited Monte Carlo experiments show that the proposed algorithm is second-order optimal for a large class of channels.
01 Jan 2001
TL;DR: The problem of transmitting a single block of data over an unknown multipath channel is considered, and it is proved that the transmitted signal should look white if the channel parameters are unknown, and this can be essentially achieved by using a zero padding precoder.
Abstract: The problem of transmitting a single block of data over an unknown multipath channel is considered. It is explained why the transmitted signal should look white if the channel parameters are unknown, and it is proved that this can be essentially achieved by using a zero padding precoder. A novel canonical representation for linear precoders is then used to decompose this zero padding precoder into a form better suited to multipath channels. Adding a cyclic prefix to this precoder results in what is called a "spectrally balanced OFDM precoder".