Showing papers on "Kernel adaptive filter published in 1993"

A stochastic gradient adaptive filter with gradient adaptive step size

Abstract: The step size of this adaptive filter is changed according to a gradient descent algorithm designed to reduce the squared estimation error during each iteration. An approximate analysis of the performance of the adaptive filter when its inputs are zero mean, white, and Gaussian noise and the set of optimal coefficients are time varying according to a random-walk model is presented. The algorithm has very good convergence speed and low steady-state misadjustment. The tracking performance of these algorithms in nonstationary environments is relatively insensitive to the choice of the parameters of the adaptive filter and is very close to the best possible performance of the least mean square (LMS) algorithm for a large range of values of the step size of the adaptation algorithm. Several simulation examples demonstrating the good properties of the adaptive filters as well as verifying the analytical results are also presented. >

Perfect reconstruction filter banks with rational sampling factors

Abstract: An open problem, namely, how to construct perfect reconstruction filter banks with rational sampling factors, is solved. Such filter banks have N branches, each one having a sampling factor of p i/qi, and their sum equals one. In this way, the well-known theory of filter banks with uniform band splitting is extended to allow for nonuniform divisions of the spectrum. This can be very useful in the analysis of speech and music. The theory relies on two transforms. The first transform leads to uniform filter banks having polyphase components as individual filters. The other results in a uniform filter bank containing shifted versions of same filters. This, in turn, introduces dependencies in design, and is left for future work. As an illustration, several design examples for the (2/3, 1/3) case are given. Filter banks are then classified according to the possible ways in which they can be built. It is shown that some cases cannot be solved even with ideal filters (with real coefficients)

Auditory prosthesis, noise suppression apparatus and feedback suppression apparatus having focused adaptive filtering

Abstract: A noise and feedback suppression apparatus processes an audio input signal having both a desired component and an undesired component. When implemented so as to effect noise cancellation, the apparatus includes a first filter operatively coupled to the input signal. The first filter generates a focused reference signal by selectively passing an audio spectrum of the input signal which primarily contains the undesired component. The reference signal is supplied to an adaptive filter disposed to filter the input signal so as to provide an adaptive filter output signal. A combining network subtracts the adaptive filter output signal from the input signal to create an error signal. The noise suppression apparatus further includes a second filter for selectively passing to the adaptive filter an audio spectrum of the error signal substantially encompassing the spectrum of the undesired component of the input signal. This cancellation effectively removes the undesired component from the input signal without substantially affecting the desired component of the input signal. When the present apparatus is implemented so as to suppress feedback the adaptive filter output signal is employed to cancel a feedback component from the input signal.

Fuzzy adaptive filters, with application to nonlinear channel equalization

Abstract: Two fuzzy adaptive filters are developed: one uses a recursive-least-squares (RLS) adaptation algorithm, and the other uses a least-mean-square (LMS) adaptation algorithm. The RLS fuzzy adaptive filter is constructed through the following four steps: (1) define fuzzy sets in the filter input space Rn whose membership functions cover U; (2) construct a set of fuzzy IF-THEN rules which either come from human experts or are determined during the adaptation procedure by matching input-output data pairs; (3) construct a filter based on the set of rules; and (4) update the free parameters of the filter using the RLS algorithm. The design procedure for the LMS fuzzy adaptive filter is similar. The most important advantage of the fuzzy adaptive filters is that linguistic information (in the form of fuzzy IF-THEN rules) and numerical information (in the form of input-output pairs) can be combined in the filters in a uniform fashion. The filters are applied to nonlinear communication channel equalization problems. >

The gamma-filter-a new class of adaptive IIR filters with restricted feedback

Abstract: Generalized feedforward filters, a class of adaptive filters that combines attractive properties of finite impulse response (FIR) filters with some of the power of infinite impulse response (IIR) filters, are described. A particular case, the gamma filter, generalizes Widrow's adaptive transversal filter (adaline) to an infinite impulse response filter. Yet, the stability condition for the gamma filter is trivial, and LMS adaptation is of the same computational complexity as the conventional transversal filter structure. Preliminary results indicate that the gamma filter is more efficient than the adaptive transversal filter. The authors extend the Wiener-Kopf equation to the gamma filter and develop some analysis tools. >

Effects of Multirate Systems on the Statistical Properties of Random Signals

Abstract: In multirate digital signal processing, we often encounter time-varying linear systems such as decimators, interpolators, and modulators. In many applications, these building blocks are interconnected with linear filters to form more complicated systems. It is often necessary to understand the way in which the statistical behavior of a signal changes as it passes through such systems. While some issues in this context have an obvious answer, the analysis becomes more involved with complicated interconnections. For example, consider this question: if we pass a cyclostationary signal with period K through a fractional sampling rate-changing device (implemented with an interpolator, a nonideal low-pass filter and a decimator), what can we say about the statistical properties of the output? How does the behavior change if the filter is replaced by an ideal low-pass filter? In this paper, we answer questions of this nature. As an application, we consider a new adaptive filtering structure, which is well suited for the identification of band-limited channels. This structure exploits the band-limited nature of the channel, and embeds the adaptive filter into a multirate system. The advantages are that the adaptive filter has a smaller length, and the adaptation as well as the filtering are performed at a lower rate. Using the theory developed in this paper, we show that a matrix adaptive filter (dimension determined by the decimator and interpolator) gives better performance in terms of lower error energy at convergence than a traditional adaptive filter. Even though matrix adaptive filters are, in general, computationally more expensive, they offer a performance bound that can be used as a yardstick to judge more practical "scalar multirate adaptation" schemes.

A fast recursive least squares adaptive second order Volterra filter and its performance analysis

Abstract: A fast, recursive least squares (RLS) adaptive nonlinear filter modeled using a second-order Volterra series expansion is presented. The structure uses the ideas of fast RLS multichannel filters, and has a computational complexity of O(N/sup 3/) multiplications, where N-1 represents the memory span in number of samples of the nonlinear system model. A theoretical performance analysis of its steady-state behaviour in both stationary and nonstationary environments is presented. The analysis shows that, when the input is zero mean and Gaussian distributed, and the adaptive filter is operating in a stationary environment, the steady-state excess mean-squared error due to the coefficient noise vector is independent of the statistics of the input signal. The results of several simulation experiments show that the filter performs well in a variety of situations. The steady-state behaviour predicted by the analysis is in very good agreement with the experimental results. >

Applying an adaptive feed-forward algorithm as a frequency selective filter in a closed loop disk drive servo system in order to compensate for periodic perturbations which otherwise appear in the servo system position error signal

Abstract: A method for applying an adaptive feed-forward waveform correction to a primary servo loop compensation signal in a rotating data storage apparatus. A position error signal is used to determine a fixed feed-forward correction upon initialization or other predetermined conditions. An adaptive feed-forward waveform correction is periodically determined by adding a scaled version of the position error signal to the stored adaptive feed-forward value, and then processing the resulting sum through a frequency selective filter. The stored adaptive feed-forward correction is then updated with this new result. Simultaneously, the updated adaptive feed-forward correction is combined with the fixed feed-forward correction and the position error signal to generate a primary servo loop compensation signal for controlling the read/write head of the data storage apparatus.

Speech processing system and method for enhancing a speech signal in a noisy environment

Abstract: A speech processing system (30) operates in a noisy environment (20) by performing adaptive prediction between inputs from two sensors positioned to transduce speech from a speaker, such as an accelerometer and a microphone. An adaptive filter (37) such as a finite impulse response (FIR) filter receives a digital accelerometer input signal, adjusts filter coefficients according to an estimation error signal, and provides an enhanced speech signal as an output. The estimation error signal is a difference between a digital microphone input signal and the enhanced speech signal. In one embodiment, the adaptive filter (37) selects a maximum one of a first predicted speech signal based on a relatively-large smoothing parameter and a second predicted speech signal based on a relatively-small smoothing parameter, with which to normalize a predicted signal power. The predicted signal power is then used to adapt the filter coefficients.

A high sampling rate delayed LMS filter architecture

Abstract: The problem of implementing a high sampling rate transversal form adaptive filter is investigated. A highly pipelined systolic-type alternative to the conventional LMS adaptive filter is presented. The proposed system consists of a linear array of identical processing modules specifically suited to the computational requirements of the delayed LMS algorithm. The resulting adaptive filter structure can accommodate very high sampling rates, which are independent of the filter order. The performance of the system is analyzed in terms of computational speedup and maximum sampling rate, and the effect of adaptation delay on algorithm convergence is addressed. >

Multi-channel echo cancellation with adaptive filters having selectable coefficient vectors

Abstract: A multi-channel echo canceller comprises a set of subtractors connected respectively in communication channels for respectively receiving a signal from a respective microphone and cancelling an echo contained in it with a cancelling signal, and a set of adaptive filters associated respectively with the subtractors. Each adaptive filter has a set of vectors of filter coefficients. A time difference between propagation delays of the received signals is estimated, and a signal having the largest content of echo components is selected and applied to the adaptive filters. One of the coefficient vectors is identified according to the estimated time difference and the selected signal. Each adaptive filter varies its filter coefficients of the identified vector with a correction term proportional to the output of the associated subtractor for filtering the selected signal using the coefficients of the identified vector to derive an echo replica, which is supplied to the associated subtractor as the cancelling signal.

Design and characterization of optimal FIR filters with arbitrary phase

Abstract: An algorithm for designing a Chebyshev optimal FIR filter that approximates an arbitrary complex-valued frequency response is presented. This algorithm computes the optimal filter by solving the dual to the filter design problem. It is guaranteed to converge theoretically and requires O(N/sup 2/) computations per iteration for a filter of length N. For the first time, properties of the optimal filter are derived, and the case where the desired filter has arbitrary constant group delay is studied in detail. >

Minimum mean-square-error filter for pattern recognition with spatially disjoint signal and scene noise

Abstract: A minimum mean-square-error filter for pattern-recognition problems with input scene noise that is spatially disjoint (or nonoverlapping) with the target is described. The filter is designed to locate the target by producing a delta function output at the target position. The filter minimizes the mean square of the difference between the desired output delta function and the filter output in response to a noisy input data. We show that the filter output has a well-defined peak and small sidelobes in the presence of spatially disjoint target and scene noise.

Performance analysis of adaptive filter structures based on subband decomposition

Abstract: The properties of adaptive finite impulse response (FIR) filters in subbands are investigated, where the input signal is first decomposed in subband signals which are then modified at a lower rate by shorter adaptive filters. Two structures, with different coefficient adaptations, are studied. Expressions for the errors introduced by the nonideal characteristics of the filter banks are found, and a method for the optimal design of the analysis and synthesis banks for the adaptive filtering context is presented. The effect of the non-ideal filter bank characteristics in the adaptation speed of the multirate structures is investigated. The results of computer simulations considering an acoustic echo canceller application are presented. >

Low-delay subband adaptive filter

Abstract: A subband adaptive filter is disclosed that retains the computational and convergence speed advantages of subband processing while eliminating delay in the signal path. The technique has applications in active noise control where delay seriously limits cancellation performance and in acoustic echo cancellation where transmission delay specifications may limit the use of conventional subband designs. A reference signal and a residual error signal are each decomposed into a plurality of subband signals, and a set of adaptive weighting coefficients is generated for each of these subbands by a conventional complex LMS (least-mean-squared) technique. These sets of subband weighting coefficients are then transformed into the frequency domain, appropriately stacked and inverse transformed back into the time domain to obtain wideband filter coefficients for a programmable filter. The reference signal, which is correlated with the disturbance signal to be eliminated, is filtered by this programmable filter to produce a disturbance estimate signal which may be subtracted from the signal containing the disturbance. The process iterates in order to minimize the residual error signal and thereby reduce the disturbance.

Two-dimensional LMS adaptive filter incorporating a local-mean estimator for image processing

Abstract: An adaptive scheme is proposed whereby a local-mean estimation substructure is incorporated to overcome the problem of the 2-D least-mean-square (LMS) adaptive filter in dealing with signals having a nonzero mean, especially with abrupt variations (edges), which is a typical feature of practical images. The advantages of this scheme over the direct use of the 2-D LMS are considered. As a direct application, the scheme is used for adaptive cancellation of an artifact disturbance in an image. A method of implementing the local-mean estimation substructure with the edge-preserving property is introduced. Using this local-mean estimator, the scheme is applied to enhancement of images disturbed by wideband noise. All of the application examples illustrate some useful feature of the 2-D LMS adaptive filter in image processing and the improved performance of the scheme. >

A new adaptive separable median filter for removing blocking effects

Abstract: A novel adaptive separable median filter as a postfilter is presented for removing the blocking effects which generally originate from lower-bit-rate image transmission. This filter adaptively transforms from a traditional median filter to a low-pass filter progressively when the filter is close to the position of blocking effects. Simulation results demonstrate that this filter is extremely useful not only in removing the blocking effects, but also in maintaining the main advantages of the median filter such as edge preservation and noise reduction. >

Method of performing convergence in a, least mean square, adaptive filter, echo canceller

Abstract: A method is described of decreasing convergence time in an adaptive echo canceller (20). The method includes the step of locating a primary echo within a filter vector based upon relative tap values within the filter vector. The filter vector is then narrowed, based upon the located taps. An estimated error is determined based, in part, upon the narrowed filter vector. An updated filter vector is produced based, in part, upon the estimated error.

Image texture synthesis-by-analysis using moving-average models

Abstract: A synthesis-by-analysis model for texture replication or simulation is presented This model can closely replicate a given textured image or produce another image that although distinct from the original, has the same general visual characteristics and the same first and second-order gray-level statistics as the original image The texture synthesis algorithm, proposed contains three distinct components: a moving-average (MA) filter, a filter excitation function, and a gray-level histogram The analysis portion of the texture synthesis algorithm derives the three from a given image The synthesis portion convolves the MA filter kernel with the excitation function, adds noise, and modifies the histogram of the result The advantages of this texture model over others include conceptually and computationally simple and robust parameter estimation, inherent stability, parsimony in the number of parameters, and synthesis through convolution The authors describe a procedure for deriving the correct MA kernel using a signal enhancement algorithm, demonstrate the effectiveness of the model by using it to mimic several diverse textured images, discuss its applicability to the problem of infrared background simulation, and include detailed algorithms for the implementation of the model >

Active noise reduction system for automobile compartment

Abstract: In an automobile compartment noise reduction system, an ignition signal transforming circuit processes an ignition pulse signal to obtain a vibration noise source signal with a frequency spectrum composed of 0.5×n (integers) order components of the engine r.p.m. as the primary source signal. The signal is applied to an adaptive filter and an LMS calculating circuit via a speaker-microphone transmission characteristic correcting circuit. The primary source signal is synthesized by the filter into a cancel signal and then outputted through a speaker as canceling sound. The canceling sound is received by at least one error microphone at a noise receiving point as an error signal. The error signal is applied to the LMS calculating circuit. The LMS circuit updates the filter coefficients of the adaptive filter on the basis of the primary source signal and the error signal so that the error signal can be minimized. The noise reduction system has high reliability with low cost, and is easy to mount.

Active control apparatus using adaptive digital filter

Abstract: An active control apparatus using an adaptive IIR digital filter includes a coefficient control portion for updating the filter coefficient of adaptive IIR digital filter so as to minimize an output error signal level, and a coefficient control portion for updating the filter coefficient of a recursive portion so as to minimize the output level of recursive portion. The filter coefficient of recursive filter is updated by these two coefficient control portions and in parallel or in a time-dividing manner.

Fractionally-spaced equalizer for a DS-CDMA system

Abstract: An adaptively-tuned filter (80) for use within the receiver (14) of a multiple-access code division (CDMA) communication system permits spectrally-inefficient transmissions to be made from each transmitter (12) used with such system, thereby simplifying the transmitter circuits, yet still achieves a signal-to-noise ratio (SNR) that approaches the SNR achieved when spectrally-efficient transmissions are made. The adaptive filter (80) also compensates for signal distortions resulting from multiple signal paths. The receiver (14) includes RF receiving circuits (60, 62), a matched filter (64), downconversion circuits (66, 68, 92), sampling circuits (70, 71, 72), a time acquisition circuit (76), an adaptive filter (80), a decimator circuit (75), a despreader circuit (77), and an accumulator circuit (78). The adaptive filter (80) includes, e.g., two LMS gradient filters (96a, 96b), or equivalent, that operate independently in parallel, with an accurate bit decision being used to select the tap signals to be used as the starting point for the next bit. The filters (80, 96) function as a fractionally-spaced chip-MSE equalizer that improves the bit-SNR in the presence of sufficient multi-access noise, and also provides narrowband noise rejection.

Exact reconstruction analysis/synthesis filter banks with time-varying filters

Abstract: The authors examine some of the analysis/synthesis issues associated with FIR (finite impulse response) time-varying filter banks where the filter bank coefficients are allowed to change in response to the input signal. Several issues are identified as being important for realizing performance gains from time-varying filter banks in image coding applications. These issues relate to the behavior of the filters as transition from one set of filter banks to another occurs. Lattice structure formulations for the time-varying filter bank problems are introduced and discussed in terms of their properties and transition characteristics. >

Adaptive switched-capacitor filters based on the LMS algorithm

Abstract: The implementation of adaptive FIR filters based on the LMS algorithm using SC circuits is described. Basically, the filters consist of a delay element, a summing circuit, an integrator, and a multiplier. The influence of nonideal effects of SC networks on the behavior of a given filter is investigated. It is shown that the nonidealities in the FIR filter part of the circuit can be eliminated by an additional constant tap element, whereas the main error source in the adaptation part is the multiplier- and integrator offset-errors. The errors can be compensated for using special offset-free circuits. Using the proposed offset-compensation schemes, the accuracy of a switched-capacitor adaptive filter is mainly determined by the nonlinearity errors of the multipliers. >

Adaptive lattice bilinear filters

Abstract: Two fast least-squares lattice algorithms for adaptive nonlinear filters equipped with bilinear system models are presented. The lattice filter formulation transforms the nonlinear filtering problem into an equivalent multichannel linear filtering problem, thus using multichannel lattice filtering algorithms to solve the nonlinear filtering problem. The computational complexity of the algorithms is an order of magnitude smaller than that of previously available methods. The first of the two approaches is an equation error algorithm that uses the measured desired response signal directly to compute the adaptive filter outputs. This method is conceptually very simple, but results in biased system models in the presence of measurement noise. The second is an approximate least-squares output error solution; the past samples of the output of the adaptive system itself are used to produce the filter output at the current time. Results indicate that the output error algorithm is less sensitive to output measurement noise than the equation error method. >

Adaptative equalizing receiver and maximum likelihood sequence estimation receiver

Abstract: A receiver for compensating the deterioration of transmission characteristics, in which a transmission path estimation circuit (5) estimates the impulse response of the transmission path from the received signal. Bases upon the estimated impulse response, a switch (7) outputs selectively one of the outputs of an adaptative equalizing circuit (10) which adaptively equalize the received signal, and of a sign judging circuit (9) which is output only by judging the sign of the received signal. Then, a variable number-of-tap adaptive filter is used for the adaptive equalization of the received signal. Based on the estimated impulse response, the number-of-tap setting means controls the number of taps of the adaptive filter. Also, for the adaptive equalization of the received signal, there are used a transversal type matching filter, and a status estimation circuit which estimates the maximum likelihood of the transmission symbol string by the output from the matching filter. The number-of-tap setting means controls the number of statuses to be considered in accordance with the estimated impulse response.

Line search algorithms for adaptive filtering

Abstract: Line search algorithms for adaptive filtering that choose the convergence parameter so that the updated filter vector minimizes the sum of squared errors on a linear manifold are described. A shift invariant property of the sample covariance matrix is exploited to produce an adaptive filter stochastic line search algorithm for exponentially weighted adaptive equalization requiring 3N+5 multiplications and divisions per iteration. This algorithm is found to have better numerical stability than fast transversal filter algorithms for an application requiring steady-state tracking capability similar to that of least-mean square (LMS) algorithms. The algorithm is shown to have faster initial convergence than the LMS algorithm and a well-known variable step size algorithm having similar computational complexity in an adaptive equalization experiment. >