HomeChair: Piet C.W. Sommen, Eindhoven University of Technology, Netherlands
Home
Carlos Mosquera, Universidad de Vigo (Spain)
Jose A Gomez, Universidad de Vigo (Spain)
Fernando Perez-Gonzalez, Universidad de Vigo (Spain)
This paper is concerned with systems in which the output error of an adaptive IIR filter is subsequently filtered, e.g., an active noise control system where a transfer function models the path between the injection point of the cancellation noise and the point where the residual error is measured. We propose a family of algorithms suited to this type of scenarios, deriving conditions for their deterministic convergence. The analysis of the convergence is particularized to the Filtered-U Recursive LMS algorithm, a popular scheme whose global convergence has never been proved formally. Finally, some results based on real measurements are also presented.
ic981277.pdf (From Postscript)
TOPTõnu Trump, Ericsson Radio Systems AB (Sweden)
The problem of incorporating partial knowledge of measurement noise into a frequency domain adaptive filtering scheme is addressed.The proposed algorithm is obtained by minimizing a BLUE criterion function using the stochastic gradient method and then switching over to the frequency domain to reduce the computational complexity. The performance of the algorithm in the situations of colored measurement noise is demonstrated by means of simulations using stationary as well as speech signals.
ic981286.pdf (From Postscript) TOP
Salvador Olmos, University of Zaragoza (Spain)
Jose Garcia, University of Zaragoza (Spain)
Raimon Jane, Politechnic University of Catalonia (Spain)
Pablo Laguna, University of Zaragoza (Spain)
In this work we show that orthogonal expansions of recurrent signals like electrocardiograms with a reduced number of coefficients can be considereded as a periodic time-variant filter. Instantaneous impulse and frequency responses are analyzed for two cases: estimation of the coefficients with inner product and adaptive estimation with LMS algorithm.
ic981559.pdf (From Postscript) TOP
Youhua Wang, Kanazawa University (Japan)
Kazushi Ikeda, Kanazawa University (Japan)
Kenji Nakayama, Kanazawa University (Japan)
The numerical property of an adaptive filter algorithm is the most important problem in practical applications. Most fast adaptive filter algorithms have the numerical instability problem and the fast Newton transversal filter (FNTF) algorithms are no exception. In this paper, we propose a numerically stable fast Newton type adaptive the proposed algorithm from the order update fast least squares (FLS) algorithm. This derivation is direct and simple to understand. Second, we give the stability analysis using linear time-variant state-space method. The transition matrix of the proposed algorithm is given. Theeigenvalues of the ensemble average of the transition matrix are shown to be asymptotically all less than unity. This results in a much improved numerical performance compared with the FNTF algorithms. The computer simulations implemented by using a finite-precision arithmetic have confirmed the validity of our analysis.
ic981919.pdf (Scanned) TOPThomas Schertler, Darmstadt University of Technology (Germany)
Adaptive filters for the cancellation of acoustic echoes, as applied in hands-free telephone sets, require about a thousand coefficients and more to get a significant echo reduction. This leads to a very high computational effort and cannot be realized on most low-cost DSPs. One common proposition to decrease the computational load is to update only a portion of the coefficients at a time. This decreases not only the computational load but also the convergence speed. To reduce this drawback, it has been suggested that only the most significant coefficients be updated. This improves the convergence speed considerably. Unfortunately, it requires additional memory of twice the filter length. In our proposal, we present a modified version of the mentioned algorithm which has almost the same adaptation speed but consumes significantly less memory.
ic982075.pdf (From Postscript) TOPTony Gustafsson, Chalmers University of Technology (Sweden)
In this paper an instrumental variable (IV) based subspace tracking algorithm is proposed. The basic idea of the algorithm is to reduce the amount of computations using a certain perturbation/approximation strategy. The complexity is reduced to mnn, which should be compared to mll for the SVD, where m,l >> n in general (mdenotes the number of sensors, l denotes the number of instruments, and n denotes the number of signals). The proposed algorithm turns out to be related to Karasalo's subspace averaging approach. In a series of simulations we demonstrate that the detection-, stationary estimation-, and tracking performance of the proposed algorithm is essentially equivalent to that achieved by the truncated SVD.
ic981580.pdf (From Postscript) TOP
James R Saffle, Villanova University (U.S.A.)
Sathyanarayan S Rao, Villanova University (U.S.A.)
A closed-loop adaptive subband prediction architecture is presented by employing an adaptive subband filter in the prediction configuration. Some authors have suggested that applying open-loop prediction methods to subband signals can realize increased prediction gain over fullband prediction. Furthermore, the benefits of applying multirate techniques to adaptive filtering are well understood in terms of reduction of computational complexity and increased convergence speed. Thus, the closed-loop subband adaptive predictor is a novel approach that is expected to exhibit these same benefits along with the advantages of backward adaptation. Results show that the new subband predictor can produce a higher prediction gain than a similar fullband adaptive prediction filter. The proposed architecture is implemented in C++ on the Pentium processor.
ic981527.pdf (From Postscript) TOP
K. Mayyas, University of Science and Technology (Jordan)
T. Aboulnasr, University of Ottawa (Canada)
A modification to the OE IIR system structure is proposed to ensure global convergence for sufficient modeling of the unknown system. The proposed structure is effectively equivalent to whitening the input signal before being applied to the original OE setup. This guarantees the unimodality of the error surface for sufficient modeling. An adaptive update scheme for the new structure is derived based on the least mean square (LMS) technique. Examples are provided to demonstrate the effectiveness of the proposed structure under different conditions.
ic981726.pdf (Scanned) TOP
Emrah Acar, Carnegie Mellon University (U.S.A.)
Orhan Arikan, Bilkent University (Turkey)
To take advantage of fast converging multi-channel recursive least squares algorithms, we propose an adaptive IIR system structure consisting of two parts: a two-channel FIR adaptive filter whose parameters are updated by rotation-based multi-channel least squares lattice (QR-MLSL) algorithm, and an adaptive regressor which provides more reliable estimates to the original system output based on previous values of the adaptive system output and noisy observation of the original system output. Two different regressors are investigated and robust ways of adaptation of the regressor parameters are proposed. Based on extensive set of simulations, it is shown that the proposed algorithms converge faster to more reliable parameter estimates than LMS type algorithms.
ic981791.pdf (From Postscript) TOP
Linshan Li, Northwestern Polytechnical University (China)
V. John Mathews, University of Utah (U.S.A.)
Parallel-cascade realizations of truncated Volterra systems implement higher-order systems using a parallel connection of multiplicative combinations of lower-order systems. Such realizations are modular and permit efficient approximations of truncated Volterra systems. Frequency-domain realizations of the least-mean-square (LMS) adaptive filter and the normalized LMS adaptive filter that implements the system model using the parallel-cascade structure are presented in this paper. Computational complexity analysis and simulation results show that the normalized frequency-domain, parallel-cascade LMS adaptive quadratic filter has the advantages of computational simplicity and superior performance over direct form realizations.
ic982096.pdf (From Postscript) TOP