Results 1 
9 of
9
Sparse LMS for system identification
 in Proc. IEEE ICASSP
, 2009
"... We propose a new approach to adaptive system identification when the system model is sparse. The approach applies the ℓ1 relaxation, common in compressive sensing, to improve the performance of LMStype adaptive methods. This results in two new algorithms, the ZeroAttracting LMS (ZALMS) and the Re ..."
Abstract

Cited by 42 (6 self)
 Add to MetaCart
(Show Context)
We propose a new approach to adaptive system identification when the system model is sparse. The approach applies the ℓ1 relaxation, common in compressive sensing, to improve the performance of LMStype adaptive methods. This results in two new algorithms, the ZeroAttracting LMS (ZALMS) and the Reweighted ZeroAttracting LMS (RZALMS). The ZALMS is derived via combining a ℓ1 norm penalty on the coefficients into the quadratic LMS cost function, which generates a zero attractor in the LMS iteration. The zero attractor promotes sparsity in taps during the filtering process, and therefore accelerates convergence when identifying sparse systems. We prove that the ZALMS can achieve lower mean square error than the standard LMS. To further improve the filtering performance, the RZALMS is developed using a reweighted zero attractor. The performance of the RZALMS is superior to that of the ZALMS numerically. Experiments demonstrate the advantages of the proposed filters in both convergence rate and steadystate behaviors under sparsity assumptions on the true coefficient vector. The RZALMS is also shown to be robust when the number of nonzero taps increases. Index Terms — LMS, compressive sensing, sparse models, zeroattracting, l1 norm relaxation 1.
Linear Regression With a Sparse Parameter Vector
 in IEEE Percentage of correctly selected order BOSS, σ 2 = −10 dB BOSS empirical AIC c BIC −5 noise variance σ 2
, 2007
"... We consider linear regression under a model where the parameter vector is known to be sparse. Using a Bayesian framework, we derive a computationally efficient approximation to the minimum meansquare error (MMSE) estimate of the parameter vector. The performance of the soobtained estimate is illus ..."
Abstract

Cited by 36 (5 self)
 Add to MetaCart
(Show Context)
We consider linear regression under a model where the parameter vector is known to be sparse. Using a Bayesian framework, we derive a computationally efficient approximation to the minimum meansquare error (MMSE) estimate of the parameter vector. The performance of the soobtained estimate is illustrated via numerical examples.
Sparse channel estimation with zero tap detection
 IEEE Trans. Wireless Commun
, 2007
"... Algorithms for the estimation of a channel whose impulse response is characterized by a large number of zero tap coefficients are developed and compared. Exploiting the sparsity of the channel, the estimation problem is transformed into an equivalent onoff keying (OOK) detection problem, whose solu ..."
Abstract

Cited by 30 (0 self)
 Add to MetaCart
(Show Context)
Algorithms for the estimation of a channel whose impulse response is characterized by a large number of zero tap coefficients are developed and compared. Exploiting the sparsity of the channel, the estimation problem is transformed into an equivalent onoff keying (OOK) detection problem, whose solution gives an indication on the position of the zero taps. The proposed schemes are compared to the standard least squares estimate (LSE) via simulations in terms of mean square error (MSE) and bit error rate (BER). A scheme based on sphere decoding appears to give the best performance while maintaining moderate complexity. 1
Dynamic Resource Allocation For Network Echo Cancellation
, 2001
"... Network echo canceler chips are designed to handle several channels simultaneously. With the processing speeds now available, a single chip might handle several hundred channels. In current implementations, however, the adaptation algorithm is designed for a single channel, and the computations are ..."
Abstract

Cited by 2 (0 self)
 Add to MetaCart
Network echo canceler chips are designed to handle several channels simultaneously. With the processing speeds now available, a single chip might handle several hundred channels. In current implementations, however, the adaptation algorithm is designed for a single channel, and the computations are replicated N c times, where Nc is the number of channels. With such an implementation, the computational requirement is N c times the peak load for a single channel. The number of computations required in each channel, however, varies widely over time. Therefore, a considerable reduction in computational load can be achieved by designing the system for the average load plus a margin to account for load variations. The reduction in complexity is achieved by exploiting three features: (a) the inherent pauses in conversations, (b) the sparseness of network echo paths, and (c) the fact that an adaptive filter does not need to be updated when the error signal is small. In this paper it is shown that, in principle, such a design can reduce the computational load by a very large factor  perhaps as large as thirty. It remains to be seen whether a customized hardware architecture can be implemented to fully take advantage of the proposed algorithm.
Prediciton of Mobile Radio Channels
, 2000
"... To whom it may concern iv Abstract Prediction of the coefficients of mobile radio channel is of interest for a range of applications such as power control, adaptive resource allocation, adaptive coding and modulation. Power control in e.g. WCDMA requires shortterm prediction over only a small fract ..."
Abstract
 Add to MetaCart
(Show Context)
To whom it may concern iv Abstract Prediction of the coefficients of mobile radio channel is of interest for a range of applications such as power control, adaptive resource allocation, adaptive coding and modulation. Power control in e.g. WCDMA requires shortterm prediction over only a small fraction of the distance between two dips in the shortterm fading pattern. Radio resource allocation and planning would require accurate and more longterm prediction, the longer the better. The performance of different predictors for the mobile radio channel are evaluated partly on simulated data, using a spherical wave propagation model, but mainly on measured broadband channel impulse responses from a suburban environment. The focus is mainly on adaptive and nonadaptive linear FIR predictors but quadraticVolterra and MARS predictors are also studied. The received power of a mobile radio channel is predicted as the sum of the squared magnitudes of the predicted individual complex taps in the channel impulse response. The linear adaptive iterated subsampled FIR predictor generally produces excellent predictions of both complex taps and total power for short ranges, that is up to 0.1 wavelengths. The performance of power predictors is reduced markedly at ranges over half a wavelengths. The advantage over using just the average power for prediction then becomes small, so we can not claim that the investigated predictors are efficient for these prediction ranges. v Acknowledgments It is all about communication, connecting people. I would especially like to express my gratitude for the communication with and connection to the following people. My patient supervisors, Prof. Anders Ahl'en and docent Mikael Sternad for thoroughness and enthusiasm, respectively. Prof. Gernot Kubin for extending my views with good humour. All three have been highly involved in the work leading to this thesis.
Scheme for Improved Residual Echo Cancellation in Packetized Audio Transmission
"... Abstract – With the advent of packetized audio transmission, such as voice over IP (VoIP), and Voice over WLAN (VoWLAN), echo cancellation has put forth many challenging requirements. When an adaptive filter echo cancellation algorithm is used, its performance can be greatly increased, and its compl ..."
Abstract
 Add to MetaCart
Abstract – With the advent of packetized audio transmission, such as voice over IP (VoIP), and Voice over WLAN (VoWLAN), echo cancellation has put forth many challenging requirements. When an adaptive filter echo cancellation algorithm is used, its performance can be greatly increased, and its complexity can be reduced if it is only applied to the active regions in which echo exists. The proposed scheme estimates the constant delay and locates the active regions and is integrated with the Fast LMS/Newton algorithm for efficient realization of long adaptive filters. We assume that the input sequence to the adaptive filter can be modeled as an AutoRegressive (AR) process whose order may be kept much lower than the adaptive filter length. In addition, the proposed echo cancellation is further integrated with residual echo cancellation scheme based on the autoregressive (AR) analysis. Residual echo is whitened by the inverse filter using the estimated AR coefficients. Removing speech characteristics of the residual echo signal, the noise reduction system successfully reduces the power of residual echo as well as that of ambient noise. The result of this integration is a powerful echo cancellation scheme that has improved performance and higher suitability for VLSI designs providing efficient operation in packetized audio transmission, hands free telephony, and commercial mobile communications.
doi:10.1155/2007/71495 Research Article DetectionGuided Fast Affine Projection Channel Estimator for Speech Applications
"... In various adaptive estimation applications, such as acoustic echo cancellation within teleconferencing systems, the input signal is a highly correlated speech. This, in general, leads to extremely slow convergence of the NLMS adaptive FIR estimator. As a result, for such applications, the affine pr ..."
Abstract
 Add to MetaCart
(Show Context)
In various adaptive estimation applications, such as acoustic echo cancellation within teleconferencing systems, the input signal is a highly correlated speech. This, in general, leads to extremely slow convergence of the NLMS adaptive FIR estimator. As a result, for such applications, the affine projection algorithm (APA) or the lowcomplexity version, the fast affine projection (FAP) algorithm, is commonly employed instead of the NLMS algorithm. In such applications, the signal propagation channel may have a relatively lowdimensional impulse response structure, that is, the number m of active or significant taps within the (discretetime modelled) channel impulse response is much less than the overall tap length n of the channel impulse response. For such cases, we investigate the inclusion of an activeparameter detectionguided concept within the fast affine projection FIR channel estimator. Simulation results indicate that the proposed detectionguided fast affine projection channel estimator has improved convergence speed and has lead to better steadystate performance than the standard fast affine projection channel estimator, especially in the important case of highly correlated speech input signals. Copyright © 2007 Yan Wu Jennifer et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. 1.