Showing papers on "Noise measurement published in 1987"

Robust adaptive beamforming

Abstract: Adaptive beamforming algorithms can be extremely sensitive to slight errors in array characteristics. Errors which are uncorrelated from sensor to sensor pass through the beamformer like uncorrelated or spatially white noise. Hence, gain against white noise is a measure of robustness. A new algorithm is presented which includes a quadratic inequality constraint on the array gain against uncorrelated noise, while minimizing output power subject to multiple linear equality constraints. It is shown that a simple scaling of the projection of tentative weights, in the subspace orthogonal to the linear constraints, can be used to satisfy the quadratic inequality constraint. Moreover, this scaling is equivalent to a projection onto the quadratic constraint boundary so that the usual favorable properties of projection algorithms apply. This leads to a simple, effective, robust adaptive beamforming algorithm in which all constraints are satisfied exactly at each step and roundoff errors do not accumulate. The algorithm is then extended to the case of a more general quadratic constraint.

Signal Detection in Non-Gaussian Noise

Abstract: This book contains a unified treatment of a class of problems of signal detection theory. This is the detection of signals in addi tive noise which is not required to have Gaussian probability den sity functions in its statistical description. For the most part the material developed here can be classified as belonging to the gen eral body of results of parametric theory. Thus the probability density functions of the observations are assumed to be known, at least to within a finite number of unknown parameters in a known functional form. Of course the focus is on noise which is not Gaussian; results for Gaussian noise in the problems treated here become special cases. The contents also form a bridge between the classical results of signal detection in Gaussian noise and those of nonparametric and robust signal detection, which are not con sidered in this book. Three canonical problems of signal detection in additive noise are covered here. These allow between them formulation of a range of specific detection problems arising in applications such as radar and sonar, binary signaling, and pattern recognition and classification. The simplest to state and perhaps the most widely studied of all is the problem of detecting a completely known deterministic signal in noise. Also considered here is the detection random non-deterministic signal in noise. Both of these situa of a tions may arise for observation processes of the low-pass type and also for processes of the band-pass type."

Control of linear systems using generalized sampled-data hold functions

Abstract: This paper investigates the use of generalized sampled-data hold functions (GSHF) in the control of linear time-invariant systems. The idea of GSHF is to periodically sample the output of the system, and generate the control by means of a hold function applied to the resulting sequence. The hold function is chosen based on the dynamics of the system to be controlled. This method appears to have several advantages over dynamic controllers: it has the efficacy of state feedback without the requirement of state estimation; it provides the control system designer with substantially more freedom; and it requires few on-line computations. This paper focuses on four questions: pole assignment, specific behavior, noise sensitivity, and robustness. Among the problems solved are: simultaneous arbitrary pole assignment for a finite number of systems by a single GSHF controller, exact model matching, decoupling, and optimal noise rejection. Examples are given.

An eigensystem realization algorithm using data correlations (ERA/DC) for modal parameter identification

Abstract: A modification to the Eigensystem Realization Algorithm (ERA) for modal parameter identification is presented in this paper. The ERA minimum order realization approach using singular value decomposition is combined with the philosophy of the Correlation Fit method in state space form such that response data correlations rather than actual response values are used for modal parameter identification. This new method, the ERA using data correlations (ERA/DC), reduces bias errors due to noise corruption significantly without the need for model overspecification. This method is tested using simulated five-degree-of-freedom system responses corrupted by measurement noise. It is found for this case that, when model overspecification is permitted and a minimum order solution obtained via singular value truncation, the results from the two methods are of similar quality.

A spectral autocorrelation method for measurement of the fundamental frequency of noise-corrupted speech

Abstract: A method for measurement of the fundamental frequency of a voiced speech signal corrupted by high levels of additive white Gaussian noise is described. The method is based on flattening the spectrum of the signal by a bank of bandpass lifters and extracting the pitch frequency from autocorrelation functions calculated at the output of the lifters. A smoothing modified median filter is applied to the calculated pitch frequency contour to result in an improvement in the accuracy of the method. A byproduct of the pitch tracker is a voiced/ unvoiced classifier. The maximum and the variance of the autocorrelation function maxima, over the bank of lifters, serve as the basis for voiced/unvoiced classification by making use of a two-dimensional, nearest-neighbor pattern recognition approach. Results are presented for fundamental frequency measurement and voiced/unvoiced classification for several signal-to-noise ratios.

Low-frequency noise in silicon-gate metal-oxide-silicon capacitors before oxide breakdown

Abstract: Measurements of the fluctuations in the tunneling current It through a thin SiO2 insulating layer, by means of an ultralow noise measurement set, showed that after a first time interval in which its power spectral density is stationary and proportional to I2t, an on‐off modulation of It arises, just before oxide breakdown. This bistable noise seems to be related to localized phenomena controlled by trapping‐detrapping processes within the oxide. Two possible mechanisms which could give rise to this bistable noise are discussed.

Two Methods for Optimal MECG Elimination and FECG Detection from Skin Electrode Signals

Abstract: Two signal processing techniques for the suppression of the maternal ECG and simultaneously optimal detection of fetal ECG with respect to noise are presented. Both techniques are based on the singular value decomposition of a measurement matrix. Criteria are given in order to evaluate, a priori, electrode locations and sampling schemes for both methods. A fundamental difference with other methods is that the number of linearly independent FECG signals is not constrained to one. One of the presented techniques is a typical offline method. It is well suited for a large number of electrodes and large number of samples, which results in a better signal to noise ratio. The second technique is a typical on-line method. It gives fetal ECG signals within about 1 s, and is adaptive to changes of the transfer (e.g., due to fetal movement). It can be applied with a small number of electrodes (e.g., eight). It is shown that if three of these signals are from thoracic electrodes, the MECG suppression is guaranteed.

Distance measure for speech recognition based on the smoothed group delay spectrum

Abstract: We present a novel spectral distance measure based on the smoothed LPC group delay spectrum which gives a stable recognition performance under variable frequency transfer characteristics and additive noise. The weight of the n-th cepstral coefficients in our measure is given by W_{n} = n^{s}. \exp(-n^{2}/2\tau^{2}) which can be adjusted by selecting proper values of s and τ. In order to optimize the parameters of this distance measure, extensive experiments are carried out in a speaker-dependent isolated word recognition system using a standard dynamic time warping technique. The input speech data used here is a set of phonetically very similar 68 Japanese city name pairs spoken by male speakers. The experimental results show that our distance measure gives a robust recognition rate in spite of the variation in frequency characteristics and signal to noise ratio(SNR). In noisy situations of segmental SNR 20 dB, the recognition rate was more than 13% higher than that obtained by using the standard Euclidean cepstral distance measure. Finally, it is shown that the optimum value of s is approximately 1, and the optimum range of τΔT is about 1 ms.

Spectral slope distance measures with linear prediction analysis for word recognition in noise

Abstract: This paper discusses the approximation and use of spectral slope distance measures derived from linear prediction analysis models of speech, with emphasis on their application for recognition of noisy speech Initial testing of these slope-based measures for speaker-dependent isolated word recognition indicates that they give considerable performance improvement over the standard cepstral distance measure in several noise conditions Comparisons are also made to two related distance measures which have been recently reported by other researchers

Alpha-Beta Filter with Correlated Measurement Noise

Abstract: For many tracking applications, the measurement errors onsuccessive observations are correlated. Using a first-order Markov model for the correlation, we present analytical expressions for the time-varying covariance and gains of an alpha-beta tracking filter.To a good approximation, the effect of correlation is to increase the time interval between measurements by a factor (1+a)/(1-a),where a is the coefficient of correlation between successive measurements.

Noise compensation in speech recognition

Abstract: In speech recognition it is advantageous to take account of noise levels both in recognition and training. In both processes signals reaching a microphone l0 are digitised and passed through a filter bank to be separated into frequency channels. In training, a noise estimator 20 and a masker l5 are used with a recognizer l8 to prepare and store probability density functions (p.d.f.s) for each channel partially defining Markov models of words to be recognized. The p.d.f.s are derived only from input signals above noise levels but derivation is such that the whole of each p.d.f. is represented. In recognition, "distance" measurements on which recognition is based are derived for each channel. If the signal in a channel is above noise then the distance is determined. by the recognizer, from the negative logarithm of the p.d.f. but if a channel signal is below noise then the distance is determined from the negative logarithm of the cumulative distance of the p.d.f. to the noise level.

Putting the visual system noise back in the picture

Abstract: Computable expressions for the input-picture-equivalent contrast noise of the visual system are provided for the locally linear subclass of nonlinear models, where the internal model noise is allowed to be signal dependent. The equivalent-noise concept is thereby extended to many of the models developed to explain masking and discrimination among suprathreshold stimuli. For these models the equivalent noise depends on the masking stimulus, and its structure can be strongly determined by the representation of the masker at the level of the system at which the performance-limiting noise is generated. The expressions are applicable to the case of less-than-full-rank transformations. Pictures that have hypothetical visual-system noise projected back into them can provide insights into efficient picture-coding algorithms.

The Relationship Between Resonator and Oscillator Noise, and Resonator Noise Measurement Techniques

Abstract: Flicker, or l/f noise in just one component, the acoustic resonator, is the dominant source of frequency fluctuations in a well designed oscillator using a bulk acoustic wave or surface acoustic wave device. Measurement of resonator noise allows prediction of oscillator noise and performance improvements through selection of devices. Leesons model for oscillator noise is reviewed, then modified to incorporate the effect of resonator noise. Typical oscillator spectra are examined using the model. Two models for resonator noise are discussed, and data presented demonstrating that the cause of resonator instability is center frequency fluctuations, not phase fluctuations. The effect of loaded Q on oscillator and resonator noise is discussed. Three means to measure resonator noise are reviewed, these being, installing the resonator in an oscillator, a dual resonator bridge, and a single resonator bridge. Calibration of each method is discussed and a simple calibration method for the single resonator bridge, which lends itself to automated testing, is proposed. Noise floor limitations of the three methods are analyzed, and typical levels to be expected are given. A correlation technique to further improve the noise floor is suggested. An improved single resonator bridge which automatically sets and holds quadrature for fast, temperature stable measurements is presented. Use of a commercial phase noise measurement system to improve accuracy and ease calibration is discussed.

Acoustic Mode Measurements in the Inlet of a Turbofan Engine

Abstract: Turbomachinery noise propagates in aircraft jet engine ducts in a complicated manner. The measurement of this propagation is useful both to identify source mechanisms and to design efficient linings. A practical method of making these measurements has been developed using linear arrays of equally spaced microphones mounted flush with the duct wall. Circumferential or axial arrays are analyzed by spatial Fourier transform, giving the sound level as a function of the spinning order or axial wavenumber, respectively. Complex demodulation is used to acquire data in a modest bandwidth around a high frequency of interest. A joint NASA/Boeing test of the system used 32 microphones in a JT15D turbofan engine inlet. A 400-Hz bandwidth centered at blade passage frequency and at one-half blade passage frequency was studied. The theoretically predicted modes were clearly seen at blade passage frequency; broadband noise at one-half blade passage frequency was biased toward modes corotating with the fan. Interference between similar modes was not a significant problem. A lining design study indicated that a 15% improvement in lining efficiency was possible for this particular engine when mode data were used. The technique has proved reliable and useful for source diagnostics and lining design.

Robustness against noise: The role of timing-synchrony measurement

Abstract: In a previous report (Ghitza, 1987, [1]) we described a computational model based upon the temporal characteristics of the information in the auditory nerve fiber firing patterns, which produced an "auditory" spectral representation (the EIH) of the input signal. We also demonstrated that for speech recognition purposes, the EIH is more robust against noise compared to the traditional Fourier power spectrum. This paper reports on the first step towards understanding the role of different parameters in the EIH in achieving this performance. Both, the Fourier power spectrum measurement and the EIH measurement can be partitioned into two parts, a filter-bank followed by feature analyzer. In the Fourier power spectrum, the filter bank consists of uniformly shaped Hamming filters and the analyzer is based on power measurements. In the EIH, the filter bank consists of the cochlear filters and the analyzer is based on timing-synchrony measurements. The present study examines the relative importance of the filter-bank properties as compared to the analysis principle. For this purpose a modified EIH model has been created in which the cochlear filters have been replaced by the uniformly shaped Hamming filters. The output of the filter bank is processed by the timing-synchrony analyzer, as with the original EIH. The modified EIH and the Fourier power spectrum differs, therefor, only in the kind of analysis performed on the filter bank output. The modified EIH has been used as a front-end to a Dynamic Time Warp (DTW), using the same set-up as in Ghitza, 1987, [1]. A speaker dependent, isolated word recognition test has been conducted, on a database consisted of a 39 word alpha-digits vocabulary spoken by two male and two female speakers, in different levels of additive white noise. Compared to the Fourier-based front-end, the recognition scores have been slightly improved in clean environment but significantly improved in noisy environments. Furthermore, compared to the original EIH, the recognition scores have also been improved, both in clean and in noisy environments. These results demonstrate that the timing-synchrony measurement is significantly more robust against noise compared to the power measurement. They also show that the robustness is due to the timing-synchrony analyzer and not to the unique shape of the cochlear filters.

Optimum Steady State Position, Velocity, and Acceleration Estimation Using Noisy Sampled Position Data

Abstract: A one-dimensional tracking filter based on the Kalman filtering techniques for tracking of a dynamic target such as an aircraft is discussed. The target is assumed to be moving with constant acceleration and is acted upon by a plant noise which perturbs its constant acceleration motion. The plant noise accounts for maneuvers and/or other random factors. Analytical results for estimating optimum steady state position, velocity, and acceleration of the target are obtained.

The role of helicopter noise‐induced vibration and rattle in human response

Abstract: Our understanding of community reaction to helicopter noise is incomplete and inadequate. While A‐weighting appears to work outdoors and at modest noise levels, and the community response in terms of percentage of population highly annoyed can be correlated with respect to the day/night average sound level (DNL) descriptor, questions remain as to the role of perceived building vibrations and rattle in human response to helicopter noise. Does hearing windows, ceiling tiles, or objects in the room rattle or does the general perception of building vibration increase the public’s adverse response to helicopter noise? To answer these questions, this study examined the role of vibration and rattle in human response to helicopter noise. Results showed that the A‐frequency‐weighting is generally adequate to assess community response to helicopter noise when no vibration or rattle is induced by the noise. When rattle or vibrations are induced by the helicopter noise, however, A‐weighting does not assess the commun...

Excess noise in dc SQUIDs from 4.2K to 0.022K

Abstract: Four types of excess noise have been identified in dc SQUIDs operated in the temperature range 0.022 to 4.2 K. At temperatures between about 2 and 4 K, the spectral density of the low frequency flux noise of a wide variety of thin-film dc SQUIDs scales as 1/fmwhere m = 1.0 ± 0.1. In SQUIDs with Nb loops the noise originates as an "apparent flux noise", whereas in those with Pb or PbIn loops the noise is substantially lower and originates in critical current fluctuations. When any of these devices is cooled to temperatures below about 0.5 K, the spectral density of the excess flux noise scales as 1/fm, with m = 0.66 ± 0.08 in most cases, and the noise always originates as an apparent flux noise. At the lowest temperatures, the white noise saturates at an effective temperature of about 150 mK; this excess noise probably arises from self-heating in the resistance shunting each tunnel junction.

The optimal projection equations for reduced-order state estimation: The singular measurement noise case

Abstract: The optimal projection equations for reduced-order state estimation are generalized to allow for singular (i.e., colored) measurement noise. The noisy and noise-free measurements serve as inputs to dynamic and static estimators, respectively. The optimal solution is characterized by necessary conditions which involve a pair of oblique projections corresponding to reduced estimator order and singular measurement noise intensity.

A single-microphone-based self-adaptive filter of noise from speech and its performance evaluation.

Abstract: This paper discusses a single-microphone-based self-adaptive filter of environmental noise from speech. This filter, based on the work of Graupe (3) and of Graupe and Causey (4), has been incorporated in standard in-the-ear (ITE) and in behind-the-ear (BTE) hearing aids by several hearing aid manufacturers. Intelligibility tests by the authors and by independent researchers are presented in this paper to illustrate the filter's performance. Significant monosyllabic-word-list intelligibility improvements are shown in hearing-impaired and in normal-hearing subjects for virtually any environmental noise, including white noise, babble (interfering background conversations), cafeteria noise, high-frequency noise, and low-frequency noise at signal-to-noise ratios to below -20 dB.

Noise suppression during seismic exploration.

Abstract: The system for suppressing an ambient noise signal which contaminates a desired signal includes a noise pickup detector (21) for detecting the ambient noise signal. A noise discriminator (22) processes the detected noise signal into noise data. A reference noise generator (40) converts the noise data into a reference noise signal. The noise generator (40) automatically adjusts the reference noise signal to correspond with changes in the noise signal as it is being detected. A filter (24) filters the reference noise signal into an error signal which is adapted to remove the noise signal from the contaminated signal. A cross-correlator (42) cross-correlates the contaminated signal with the reference noise signal and adjusts the filter (24) to produce the required error signal.

Jitter vs. additive noise in magnetic recording: Effects on detection

Abstract: Noise on thin-film magnetic recording media is often modeled as a transition jitter, whereas particulate media noise is taken to be additive. Although these two noise mechanisms can produce similar spectra their effects on detection can be markedly different. This paper analyzes these differences for several signaling/detection schemes and concludes that where transition jitter dominates, peak detection of

ARMA Modeling using cumulant and autocorrelation statistics

Abstract: One dimensional cumulant and auto-correlation output statistics are combined to form an overdetermined system of equations whose least-squares solution yields the coefficients of an ARMA model. The driving input noise is assumed to be non-Gaussian and white. The ARMA model is allowed to be non-minimum phase and even to contain all-pass factors. The special cases of AR and MA models are also included. The overdetermined nature of the method makes the solution practical for moderate output data lengths, when additive white Gaussian noise is considered. Simulations illustrate that our approach performs very well even at low signal-to-noise ratios.

Speech recognition with a noise-adapting codebook

Abstract: Speech recognizers trained in quiet conditions but operating in noise usually have poor accuracy. This paper reports two methods for improving the accuracy of an LPC vector-quantization speech recognizer by adapting the vector codebook to noisy conditions. First, each codebook vector is changed to reflect the way people speak in noise. Second, the estimated spectrum of the background noise is added to the codebook vectors. These ideas have been tested on a total of 2400 utterances of digits recorded in a car by 4 speakers. A baseline word spotter similar to NTT's SPLIT system was modified by adapting its vector codebook to noise. This adapted codebook, when used with a new word decision criterion, yields error rates at least 4 times lower for noisy conditions. The accuracy is significantly better than without codebook adaptation techniques.

Adaptive Control for Flexible Space Structures with Measurement Noise

Abstract: This paper addresses the problem of measurement noise in the design of adaptive controllers for large space structures. It is shown by simulation that although the well known "fix" based on introducing leakage terms in the adaptive law is successful in eliminating drift in the adaptive gains, the overall control demand remains unrealistically high. In order to reduce the control demand, it is necessary to attenuate the noise by introducing filters at key positions in the adaptive configuration. However, arbitrary choice and placement of noise filters is dangerous since filter lags tend to cause instabilities in the adaptation loop. This phenomenon is verified by simulation. The main result of this paper is a general theory for introducing noise filters into the adaptive controller so as to effectively suppress measurement noise, while insuring global stability of the adaptive algorithm. This result utilizes the new concept of a "branch filter" which allows filtering of the output error without introducing phase lag into the adaptive loop. The effectiveness of the proposed methods are demonstrated by simulation on an adaptive payload articulation control for a Space Station model.

A low distortion adaptive noise cancellation structure for real time applications

Abstract: This paper describes an adaptive noise cancelling structure suitable for situations where the noise reference transducer is closely spaced relative to the primary transducer The structure is based on two LMS delay line cancellers with cross coupled feedback This structure is shown, under certain circumstances, to cancel noise with low signal distortion when the transmission paths between primary and secondary sensors have low attenuation and the primary signal is continuous The system is shown to have an enhanced performance when the primary signal is intermittent and a signal energy detector is used