Open Access. Powered by Scholars. Published by Universities.®
Articles 1 - 11 of 11
Full-Text Articles in Engineering
A Novel Normalized Cross-Correlation Based Echo-Path Change Detector, M. A. Izbal, Steven L. Grant
A Novel Normalized Cross-Correlation Based Echo-Path Change Detector, M. A. Izbal, Steven L. Grant
Electrical and Computer Engineering Faculty Research & Creative Works
A double-talk detector is used to freeze acoustic echo canceller's (AEC) filter adaptation during periods of near-end speech. Increased sensitivity towards double-talk results in declaring echo-path changes as double-talk which adversely effects the performance of an AEC as we freeze adaptation when we really need to adapt. Thus, we need an efficient and simple echo-path change detector so as to differentiate any echo-path variations from double-talk condition. In this paper, we derive a novel test statistic for echo-path change detection. The proposed decision statistic detects any echo-path variations, is normalized properly and is computationally very efficient as compared to existing …
Variable Regularized Fast Affine Projections, Steven L. Grant, Asif Iqbal Mohammad, Deepak Challa
Variable Regularized Fast Affine Projections, Steven L. Grant, Asif Iqbal Mohammad, Deepak Challa
Electrical and Computer Engineering Faculty Research & Creative Works
This paper introduces a variable regularization method for the fast affine projection algorithm (VR-FAP). It is inspired by a recently introduced technique for variable regularization of the classical, affine projection algorithm (VR-APA). In both algorithms, the regularization parameter varies as a function of the excitation, measurement noise, and residual error energies. Because of the dependence on the last parameter, VR-APA and VR-FAP demonstrate the desirable property of fast convergence (via a small regularization value) when the convergence is poor and deep convergence/immunity to measurement noise (via a large regularization value) when the convergence is good. While the regularization parameter of …
A Multiple Principal Components Based Adaptive Filter, Steven L. Grant
A Multiple Principal Components Based Adaptive Filter, Steven L. Grant
Electrical and Computer Engineering Faculty Research & Creative Works
Proportionate normalized least mean squares (PNLMS) is an adaptive filter that has been shown to provide exceptionally fast convergence and tracking when the underlying system parameters are sparse. A good example of such a system is a network echo canceller. Principal components based PNLMS (PCP) extends this fast convergence property to certain nonsparse systems by applying PNLMS while using the principal components of the underlying system as basis vectors. An acoustic echo canceller is a possible example of this type of nonsparse system. Simulations of acoustic echo paths and cancellers indicate that PCP converges and tracks much faster than the …
An Improved Pnlms Algorithm, J. Benesty, Steven L. Grant
An Improved Pnlms Algorithm, J. Benesty, Steven L. Grant
Electrical and Computer Engineering Faculty Research & Creative Works
The proportionate normalized least mean square (PNLMS) algorithm was developed for use in network echo cancelers. In comparison to the normalized least mean square (NLMS) algorithm, PNLMS has a very fast initial convergence and tracking when the echo path is sparse. Unfortunately, when the impulse response is dispersive, the PNLMS converges much slower than NLMS. This implies that the rule proposed in PNLMS is far from optimal. In many simulations, it seems that we fully benefit from PNLMS only when the impulse response is close to a delta function. We propose a new rule that is more reliable than the …
Normalized Natural Gradient Adaptive Filtering For Sparse And Nonsparse Systems, Steven L. Grant, S.C. Douglas
Normalized Natural Gradient Adaptive Filtering For Sparse And Nonsparse Systems, Steven L. Grant, S.C. Douglas
Electrical and Computer Engineering Faculty Research & Creative Works
This paper introduces a class of normalized natural gradient algorithms (NNG) for adaptive filtering tasks. Natural gradient techniques are useful for generating relatively simple adaptive filtering algorithms where the space of the adaptive coefficients is curved or warped with respect to Euclidean space. The advantage of normalizing gradient adaptive filters is that constant rates of convergence for signals with wide dynamic ranges may be achieved. We show that the so-called proportionate normalized least mean squares (PNLMS) algorithm, an adaptive filter that converges quickly for sparse solutions, is in fact an NNG on a certain parameter space warping. We also show …
Dynamic Resource Allocation For Network Echo Cancellation, T. Gansler, J. Benesty, M. Mohan Sondhi, Steven L. Grant
Dynamic Resource Allocation For Network Echo Cancellation, T. Gansler, J. Benesty, M. Mohan Sondhi, Steven L. Grant
Electrical and Computer Engineering Faculty Research & Creative Works
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 N c 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 …
A Robust Proportionate Affine Projection Algorithm For Network Echo Cancellation, T. Gansler, J. Benesty, Steven L. Grant, M. M. Sondhi
A Robust Proportionate Affine Projection Algorithm For Network Echo Cancellation, T. Gansler, J. Benesty, Steven L. Grant, M. M. Sondhi
Electrical and Computer Engineering Faculty Research & Creative Works
Echo cancelers which cover longer impulse responses (greater than or equal to 64 ms) are desirable. Long responses create a need for more rapidly converging algorithms in order to meet the specifications for network echo cancelers devised by the ITU (International Telecommunication Union). In general, faster convergence implies a higher sensitivity to near-end disturbances, especially "double-talk." Previously, a fast converging algorithm called the proportionate NLMS (normalized least mean squares) algorithm (PNLMS) has been proposed. This algorithm exploits the sparseness of the echo path in order to increase the convergence rate. A robust version of PNLMS has also been presented which …
An Efficient, Fast Converging Adaptive Filter For Network Echo Cancellation, Steven L. Grant
An Efficient, Fast Converging Adaptive Filter For Network Echo Cancellation, Steven L. Grant
Electrical and Computer Engineering Faculty Research & Creative Works
This paper discusses a fast efficient adaptive filtering algorithm for network echo cancellers PNLMS++ (proportionate normalized least mean squares ++). Compared to the conventional normalized least mean squares (NLMS) algorithm, PNLMSI++ converges much more quickly when the echo path is sparse. When the echo path is dispersive, the convergence rate is the same as NLMS. In addition, the new algorithm diverges at the same rate and to the same misalignment level as NLMS during periods of undetected double-talk. PNLMS++ is only 50% more computationally complex than NLMS and requires no additional memory
Dynamically Regularized Fast Rls With Application To Echo Cancellation, Steven L. Grant
Dynamically Regularized Fast Rls With Application To Echo Cancellation, Steven L. Grant
Electrical and Computer Engineering Faculty Research & Creative Works
This paper introduces a dynamically regularized fast recursive least squares (DR-FRLS) adaptive filtering algorithm. Numerically stabilized FRLS algorithms exhibit reliable and fast convergence with low complexity even when the excitation signal is highly self-correlated. FRLS still suffers from instability, however, when the condition number of the implicit excitation sample covariance matrix is very high. DR-FRLS, overcomes this problem with a regularization process which only increases the computational complexity by 50%. The benefits of regularization include: (1) the ability to use small forgetting factors resulting in improved tracking ability and (2) better convergence over the standard regularization technique of noise injection. …
The Fast Affine Projection Algorithm, Steven L. Grant, S. Tavathia
The Fast Affine Projection Algorithm, Steven L. Grant, S. Tavathia
Electrical and Computer Engineering Faculty Research & Creative Works
This paper discusses a new adaptive filtering algorithm called fast affine projections (FAP). FAP''s key features include LMS like complexity and memory requirements (low), and RLS like convergence (fast) for the important case where the excitation signal is speech. Another of FAP''s important features is that it causes no delay in the input or output signals. In addition, the algorithm is easily regularized resulting in robust performance even for highly colored excitation signals. The combination of these features make FAP an excellent candidate for the adaptive filter in the acoustic echo cancellation problem. A simple, low complexity numerical stabilization method …
A Fast Converging, Low Complexity Adaptive Filtering Algorithm, Steven L. Grant
A Fast Converging, Low Complexity Adaptive Filtering Algorithm, Steven L. Grant
Electrical and Computer Engineering Faculty Research & Creative Works
This paper introduces a new adaptive filtering algorithm called fast affine projections (FAP). Its main attributes include RLS (recursive least squares) like convergence and tracking with NLMS (normalized least mean squares) like complexity. This mix of complexity and performance is similar to the recently introduced fast Newton transversal filter (FNTF) algorithm. While FAP shares some similar properties with FNTF it is derived from a different perspective, namely the generalization of the affine projection interpretation of NLMS. FAP relies on a sliding windowed fast RLS (FRLS) algorithm to generate forward and backward prediction vectors and expected prediction error energies. Since sliding …