Showing papers by "Anthony G. Constantinides published in 2002"

Parallel multistage detection for multiple antenna wireless systems

Abstract: A parallel detection method is proposed for use in the context of uncoded multiple antenna communication. The method does not require power reordering and helps reduce the system delay at the cost of higher computational complexity. The scheme also shows marked improvement in performance over the classical scheme.

Blind source separation for BLAST

Abstract: We present a new blind source separation (BSS) approach for Bell Labs layered space-time (BLAST) communication system. The proposed algorithm is derived from the multimodulus algorithm (MMA) and employs the Gram-Schmidt orthogonalization procedure. The basic idea is to adjust the real and imaginary parts of the equalizer matrix separately and then project the updated parameters to the orthogonality constraints which ensure independence among the equalizer outputs. The simulation results show that the proposed algorithm has better performance compared to the multiuser kurtosis algorithm (MUK) with comparable computational complexities.

Semi-blind MIMO channel tracking using auxiliary particle filtering

Abstract: In multiple antenna communications, the impulse responses of the channels are frequently assumed to be fixed However, if multiple antennas are used in mobile terminals, this assumption is weakened In this paper, we propose using a sequential Monte Carlo method, which is robust even in non-Gaussian environments, in conjunction with current detection algorithms for tracking the channel matrix to improve the performance of multiple antenna systems under time varying channels

Channel tracking for space-time block coded systems using particle filtering

Abstract: In space-time block coded systems, the channel responses are assumed to be quasi-stationary. However, in many situations, this may not be the case and yields unsatisfactory performance. Therefore, there is a need to address the problem by tracking the channel responses in some way. We propose the use of particle filters for tracking the channel responses in space time coded systems. Compared to the case of no channel tracking, the proposed scheme improves the bit error rate (BER) of the system by at least 2 dB at higher signal-to-noise ratios (SNR).

Time-varying allpass filters using spectral-shaped noise for signal decorrelation in stereophonic acoustic echo cancellation

Abstract: In this paper, a signal decorrelation technique is presented that aims to minimise the degradation on stereo signal perception, whilst alleviating the non-uniqueness problem in stereophonic acoustic echo cancellation (SAEC). The proposed technique is based on higher-order time-varying allpass filters (HO-TV-APF) utilising spectral-shaped noise, together with the interpolation method. The test results on real speech signals indicate improved performance in both objective and subjective manners.

Fast convergent multiuser constant modulus algorithm for use in multiuser DS-CDMA environment

Abstract: In a multi-user direct sequence code division multiple access (DS-CDMA) system, retrieval of the transmitted signals with a conventional receiver is difficult when there is chip asynchronism, multi path propagation and the associated near-far problem. The results show that despite these problems an adaptive linear receiver based on a mixed cross correlation and constant modulus algorithm (CC-CMA) has the potential to retrieve all users simultaneously. In this paper, we propose several algorithms based on Quasi-Newton type algorithms. Compared with the conventional CC-CMA algorithm, the proposed algorithms achieve faster convergence at low computational cost. Simulations support the improved convergence properties of the algorithm.

Block-Shanno multi-user constant modulus algorithm for use in multi-user CDMA environment

Abstract: In a multi-user direct sequence code division multiple access (DS-CDMA) system, retrieval of transmitted signals with a conventional receiver is difficult when there is chip asynchronism, multipath propagation and the associated near-far problem. Results presented show that despite these problems an adaptive linear receiver based on a mixed cross correlation and constant modulus algorithm (CC-CMA) has the potential to retrieve all users simultaneously. Compared with the conventional CC-CMA algorithm, the proposed algorithm achieves faster convergence. Simulations support the improved convergence properties of the algorithm.

A new family of blind CDMA receivers based on second order statistics

Abstract: In this paper, a new family of blind CDMA receivers based on second order statistics is presented. These receivers are based on alternative maximum signal to interference and noise ratio criteria involving the autocorrelation matrix of the whole or partial contribution matrix, and utilize the timing differences of the data symbol bits as well as the spreading code of the user of interest. Analyses and simulations show that the presented receivers do not suffer from the usual capacity limitation caused by the dimension (delay spread) of the multipath channel, and exhibit both higher performance and higher capacity than that of most existing second order statistics based blind receivers.

Blind adaptive CDMA receivers based on independent component separation with user-and-delay identification

Abstract: In this paper, new blind adaptive CDMA receivers based on independent component separation with user-and-delay identification are presented for practical multipath channel environments. The proposed cost functions are globally convergent and the obtained weight matrices at the global minima are in the vicinity of the inverse of the contribution matrix, to within a phase and/or permutation ambiguity. The proposed user-and-delay. identification, which is performed periodically on the obtained weight matrix, can distinguish with high probability the filter that corresponds to the desirably delayed data symbol bit of the user of interest. Analyses and simulations demonstrate that the presented blind adaptive receivers have higher user-and-delay identifiable capacity, and can achieve fast convergence speed and high steady state performance at low computational complexity.

A weighted mixed statistics algorithm for blind source separation

Abstract: Most blind source separation algorithms use either second order or higher order statistics in order to unmix the signals. In this paper we propose a novel weighted mixed statistics algorithm which performs significantly better than the single type statistics algorithms. Moreover, the algorithm is a generalisation of the single type statistics algorithm and requires thus less prior information. The weights are derived using the concept of estimating functions. Simulations are provided to show the enhanced performance of the weighted mixed statistics approach, even in mixtures were the signals contain no temporal information.

Stochastic conjugate gradient based multi-user constant modulus algorithm for use in multiuser DS-CDMA environment

Abstract: A stochastic conjugate gradient algorithm (SCGA) is proposed for the solving of the nonlinear optimization problem associated with the multiuser constant modulus algorithm (CC-CMA) for DS-CDMA receivers in a multiuser environment. The algorithm is referred to as the stochastic conjugate gradient CC-CMA (SCGCC-CMA). Simulations show that the SCGCC-CMA algorithm preserves the fast convergence rate of the block-shanno cross correlation constant modulus algorithm (BSCC-CMA), can be configured to perform similar to the recursive least squares (RLS) version of the CC-CMA algorithm, and can outperform the conventional CC-CMA for less cost. The proposed algorithms can also be used in DS/CDMA systems to solve the problem of joint blind channel equalization and blind source separation in a single-user and multiuser environment. Alternatively, a window sliding parameter may be adjusted to trade off between performance and computation to match system requirements. We also propose a convergence analysis for the proposed algorithms.

A hierarchical feedforward adaptive filter for system identification

Abstract: An architecture for adaptive filtering based upon the previously introduced hierarchical least mean square algorithm is proposed. This pyramidal architecture incorporates sparse connections between the architectural layers with a certain variable degree of overlapping between the neighboring subfilters of the same level. A learning algorithm for this class of structures is derived, based on the back-propagation algorithm for temporal feedforward networks with linear neurons. Further, a class of normalized algorithms for this class is derived. The analysis and simulations show the proposed algorithms outperform the existing ones.

Robust and fast convergent blind multi-user detectors for DS-CDMA systems in nonstationary multipath environments

Abstract: A set of adaptive receivers is proposed in this paper for direct-sequence code division multiple access (DS/CDMA) systems. Specifically we develop adaptive step-size blind algorithms which are based on constant modulus (CMA), constrained minimum output energy (CMOE) and multiuser constant modulus (MU-CMA) criteria. The applications of these algorithms to a nonstationary multipath environment were analyzed and simulation studies of the proposed algorithms were carried out. These studies show that the proposed adaptive step-size CMA, adaptive step-size CMOE and the adaptive step-size MU-CMA yield superior performance to most of the existing standard blind algorithms under severe nonstationary multipath fading channels. This is particularly the case when the number of interferers are nonstationary in the statistical sense. The robust nature of the proposed receivers is analysed also with respect to the initial step-size settings. In our simulations, both deterministic and Markovian time-varying interference statistics were considered. Specially, the Markovian time-varying case is of interest in mobile wireless systems where the number of users arrive according to a Poisson process, and hence, the interference statistics switch rapidly with time.

A novel initialization scheme for blind equalization algorithms

Abstract: A new algorithm for equalizer initialization for blind equalization algorithms is presented The proposed algorithm initially employs a heuristic approach in order to obtain an estimation of the channel's characteristics Then, through a least squares inversion, an approximation of the channel's estimate inverse is produced and is used to initialize the equalizer's tap weights The proposed algorithm exhibits superior performance compared with the center-tap initialization scheme in terms of convergence speed, shorter required equalizer lengths and lower bit error rates

Time-varying allpass filters using spectral-shaped noise for signal decorrelation in stereophonic acoustic echo cancellation

Abstract: A signal decorrelation technique is presented which aims at alleviating the non-uniqueness problem in stereophonic acoustic echo cancellation (SAEC) whilst minimising the degradation of stereo signal perception. The proposed technique is based on higher-order time-varying allpass filters (HO-TV-APFs) utilising spectral-shaped noise, together with the interpolation method. Two approaches for spectral-shaped random noises are introduced, while the test results on speech signals indicate improved performance in both objective and subjective manners.

Stochastic approximation and data re-use applied to blind adaptive algorithm for interference cancellation RAKE CDMA receiver

Abstract: This paper addresses the problem of stochastic approximation and data re-use applied to the blind adaptive algorithm for the interference cancellation RAKE CDMA receiver. The improved adaptive algorithm that utilizes step-size adaptation can achieve both faster convergence and higher steady state performance compared to the fixed step-size ones without increasing much more complexity. Moreover, the improved adaptive algorithm that further utilizes solution averaging and data re-use can further improve both convergence speed and steady state performance. Simulations support the resulting significant improvements.

Blind Source Separation with the weighted mixed statistics algorithm

Abstract: Depending on the character of the signals, most Blind Source Separation algorithms exploit either the second order or fourth order statistics of the signals. In this paper we present a novel weighted mixed statistics algorithm which performs significantly better than the single type statistics algorithms. As the algorithm is a generalisation of the single type statistics algorithm, it requires less prior information. Estimating functions are used in order to derive the weights. We provide simulations to show the enhanced performance of the weighted mixed statistics approach, even in mixtures were the signals contain no temporal information.

Blind signal recovery using a novel system identification approach

Abstract: The blind equalization approaches that have been developed and employed in the literature, in relation to moving average (MA) systems give encouraging results provided that the length of the equalizer is significantly greater than that of the channel. Motivated from a desire to reduce this length, the problem is cast in the context of system identification, in which case any length for the adaptive filter that is equal to or greater than that of the original system may be assumed. To implement the identification problem, a novel low order block based cost function of the constant modulus type is formulated, which is minimized using a zero tolerance optimization approach.

Adaptive step-size blind multi-user detectors for DS-CDMA systems under deterministically and Markovian time-varying multipath environments

Abstract: A set of adaptive receivers are presented in this paper for direct-sequence code division multiple access (DS/CDMA) systems. Specifically we develop adaptive step-size blind algorithms which are based on constant modulus (CMA), constrained minimum output energy (CMOE) and multiuser constant modulus (MU-CMA) criteria. The applications of these algorithms to a nonstationary multipath environment were analyzed and simulation studies of the proposed algorithms were carried out These studies show that the proposed adaptive step-size CMA, adaptive step-size CMOE and the adaptive step-size MU-CMA yield superior performance to most of the existing standard blind algorithms under severe nonstationary multipath fading channel. This is particularly the case when the number of interferers are nonstationary in the statistical sense. The robust nature of the proposed receivers are analysed also with respect to the initial step-size settings. In our simulations, both deterministic and Markovian time-varying interference statistics were considered. Specially, the Markovian time-varying case is of interest in mobile wireless systems where the number of users arrive according to a Poisson process, and hence, the interference statistics switch rapidly with time.