Compared of Polarity Correlation Algorithm and LMS Algorithm for Flow Velocity Measurement

The transversal adaptive filter using the least mean square (LMS) algorithm of Widrow and Ropf (1976) has been widely used mainly due to its relative ease of implementation. The major drawback of this time-domain LMS (TDLMS) algorithm is that as the eigenvalue spread of the input autocorrelation matrix increases, the convergence speed of the algorithm decreases. This led to the transform domain adaptive filters where the input signals are orthogonalized. Normalized frequency domain LMS algorithms (NFDLMS) are known to be faster than the time domain implementations. However, in some implementations with low signal to return noise ratio, NFDLMS algorithms can have stability problems. The stability problem can be solved by weighting the normalization gain /spl mu/. We perform computer simulations for the telephone echo channel and show that the modified /spl mu/-weighted NFDLMS algorithm is 8 times faster than the time-domain LMS (TDLMS) algorithm and more stable than the NFDLMS algorithm over a wide range of signal to noise ratios.

Simulation of active noise reduction using least mean squares (LMS) algorithm was carried out to support the development of the ANC (active noise control) prototype, which will be applied to a four-wheeled medium class gasoline-fueled Multi-Purpose Vehicle (MPV) with 2000-2400 cc engines. In this simulation, the LMS is applied to a set of synthetic and field data, to focus on the LMS being successfully applied to all field data, ignoring the delay time between errors and references microphone, and obtaining the optimal LMS step-size and order. Based on the simulation, positive result is shown by the LMS algorithm, where the best result, the superposition of the noise and antinoise signal attenuates and reaches convergence in about 20 ms. In addition, changes in the value of LMS orders do not significantly influence LMS predictions. On the other hand, the change in the step-size value is quite influential on the prediction, where the smaller the step-size, the longer the superposition time to reach convergence.

We describe the application of a genetic algorithm (GA) to the problem of parameter optimization for an adaptive finite impulse response (FIR) filter combining genetic algorithm (GA) and least mean square (LMS) algorithm. For system identification problem, LMS algorithm computes the filter coefficients and GA search the optimal step-size adaptively. Because step-size influences on the stability and performance, so it is necessary to apply method that can control it. The simulation results of the GA were compared to the traditional LMS algorithm. We obtained that genetic algorithm was clearly superior (in accuracy) in most cases.

In the radio observation based on the large single antenna, in order to reduce the interference of navigation signal L 2 from observation of the red-shifted H I spectral line at L-band, we used the an auxiliary channel and normalized LMS algorithm, and proposed new evaluating indicators deduced by theoretical method. Finally, we verified the suppression performance and the indicators' utility by multiple simulations and statistical method.

This paper surveys recent advances related to sparse least mean square (LMS) algorithms. Since standard LMS algorithm does not take advantage of the sparsity information about the channel being estimated, various sparse LMS algorithms that are aim at outperforming standard LMS in sparse channel estimation are discussed. Sparse LMS algorithms force the solution to be sparse by introducing a sparse penalty to the standard LMS cost function. Under the reasonable conditions on the training datas and parameters, sparse LMS algorithms are shown to be mean square stable, and their mean square error performance and convergence rate are better than standard LMS algorithm. We introduce the sparse algorithms under Gaussian noises model. The simulation results presented in this work are useful in comparing sparse LMS algorithms against each other, and in comparing sparse LMS algorithms against standard LMS algorithm.

A variable step size Least Mean Square (LMS) algorithm based on the difference of the error signal (ESD-VSS) is proposed.The convergence and steady state behavior of the ESD-VSS is analyzed and the key parameters are given.Simulations are conducted and the results show that, compared with the Modified Variable Step Size (MVSS) algorithm, modified LMS algorithm and the standard LMS algorithm, the ESD-VSS provides faster convergence speed and smaller steady state misadjustment.Index Terms -LMS algorithm, variable step size, adaptive filter, MSE I . IntroductionThe LMS algorithm proposed by Widrow and Hoff has been widely used in channel equalization [1] and system identification [2], due to its robustness and simplicity.However the tradeoff between the convergence speed and the steady state misadjustment is a challenging problem.To achieve faster convergence speed and small misadjustment, a number of modified algorithms have been proposed.Ref. [3] utilized the self-correlation to control the step size, and suppress the interference of irrelevant noise effectively, but the convergence speed and the tracking capability are still expected to be improved.Ref. [4] gave a variable step size LMS algorithm using estimate signal error and thus it is of relatively faster tracking speed and smaller steady state error.However, the computational complexity of the algorithm is larger due to the tangent and exponent arithmetic, which makes it difficult to implement.

Objectives: In case of medical emergency, an ambulance uses siren (which produces noise up to 120 dB) to alert the traffic on the road to give way. However, siren noise is troublesome to the patient lying inside. This calls for noise control inside the ambulance. Methods/Statistical Analysis: The passive noise control methods are not feasible and economically inviable so active noise control (ANC) method is applied. The ANC is electro-acoustic technique in which the noise signal is acquired and noise with opposite phase is produced and overlapped on input noise to suppress the same. The weights are updated using adaptive algorithms so as to reduce the error in ANC system. In present work, least mean square (LMS) and filtered- reference least mean square (FxLMS) algorithm based ANC systems are applied and compared for ambulance siren noise reduction. MATLAB is used for simulating the ANC systems. Findings: Both the algorithms are compared on the basis of residual noise and smoothened ensemble average square error (SEASE). From the results, it is found that both algorithms LMS and FxLMS are good in noise reduction. The noise reduction for FxLMS based ANC system is 28 dB whereas it is 25 dB for LMS based ANC system, which indicates the superiority of FxLMS algorithm over LMS algorithm. The results are obtained analytically as well as indicated by power spectral density (PSD) plots of the residual noise. The comparison of SEASE for both algorithms also indicates better performance for FxLMS algorithm. The ANC system based on FxLMS algorithm is more stable as it considers the secondary path which includes practical components such as ADC, DAC, etc. Applications/Improvements: The results indicate that FxLMS algorithm based on ANC system can be implemented on the hardware using DSP processor. It can be further used for the other applications of noise reduction such as server noise, vacuum cleaner, etc. Keywords: Active Noise Control, Ambulance Siren Noise, FxLMS Algorithm, LMS

A major complaint of individuals with normal hearing and hearing impairments is a reduced ability to understand speech in a noisy environment. This paper is proposed to remove the noise from the speech signal in various real time environments and to make the hearing impaired to understand the information by transforming speech in to text format. The quality of audio signal can be improved by filtering the degraded speech signal through adaptive filters. For Noise cancellation widrow & hoff’s Least Mean Square (LMS) algorithms are being used for simplicity in implementation. The LMS algorithm was simple in it's derivation and robust in a number of applications but it also have limitation in selection of a certain values such as step size and results in computational complexity. Hence to overcome all the limitations and to present an enhanced speech signal, a novel approach named VSSNDLMS algorithm is proposed. The performance of algorithms like Variable Step Size Normalised LMS , Normalised Differential LMS with proposed VSSNDLMS with different input signals are analysed in this paper . Finally, through simulation results the proposed VSSNDLMS algorithm converges fastly with minimum mean square error and it is useful in predicting the adaptive filter performance of various algorithms and the implementation indicate the improvement in quality of the speech signal and it appears to be favourable for hearing impaired. Keywords: VSSNDLMS, LMS Algorithm, Noise Cancellation, Speech Processing, Adaptive Filter

Accurate tracking of the GPS signal in a base band correlator is critical for the positioning accuracy. However, the performance of the received signal is degraded by the presence of a multipath signal. This paper proposes the tracking of the GPS signal in the presence of the multipath using a RAKE receiver with multiple fingers. Each finger in the RAKE receiver is modelled to track the direct and respective multipath signals using the Least Mean Square (LMS) algorithm. The LMS algorithm provides an estimate of the multipath signal, and the RAKE combiner suitably combines the navigation data from the individual finger with the estimated multipath value and provides the final navigation data. A discussion on the use of the LMS for the carrier and code tracking is also presented.

For the purpose of equalization of rapidly time variant multipath channels, we derive novel adaptation schemes based on the least mean square (LMS) algorithm. In two methods proposed in this paper, instead of a coefficient vector, a coefficient matrix is prepared for a linear transversal equalizer (LTE). Based on the estimated channel impulse response, the equalizer coefficients are selected and updated by the LMS algorithm. Computer simulation results show that with a small increase of computation, the proposed methods outperform the amplitude banded LMS equalizer recently addressed in high fade rate conditions.

Due to its simplicity the adaptive Least Mean Square (LMS) algorithm is widely used in Code-Division Multiple access (CDMA) detectors. However its convergence speed is highly dependent on the eigen value spread of the input covariance matrix. For highly correlated inputs the LMS algorithm has a slow convergence which require long training sequences and therefore low transmission speeds. Another drawback of the LMS is the trade-off between convergence speed and steady-state error since both are controlled by the same parameter, the step-size. In order to eliminate these drawbacks, the class of Variable Step-Size LMS (VSSLMS) algorithms was introduced. In this paper, we study the behavior of some algorithms belonging to the class of VSSLMS for training based multiuser detection in a CDMA system. We show that the proposed Complementary Pair Variable Step-Size LMS algorithms highly increase the speed of convergence while reducing the trade-off between the convergence speed and the output error.

Least Mean Square (LMS) algorithm is undoubtedly the most resorted to algorithm in diverse fields of engineering. Due to its simplicity it has been applied to solve numerous problems including side lobe reduction in matched filters, adaptive equalization, system identification, adaptive noise cancellation etc. In this paper we present a simple architecture for the implementation of a variant of Block LMS algorithm where the weight updation and error calculation are both calculated block wise. The algorithm performs considerably well with a slight trade off in the learning curve time and misadjustment, both of which can be adjusted by varying the step size depending on the requirement. The architecture can be further modified to perform the variants of LMS algorithm such as sign-sign, signerror and sign-data algorithms. The performance of the Simplified BLMS and LMS algorithms are compared in MATLAB simulations and the hardware outputs from the FPGA are verified with the simulations.

Gain optimized cosine transform domain LMS algorithm for adaptive filtering of EEG

In this paper, a novel single-channel anti-saturation method based on LMS (Least Mean Square) algorithm and FDL (Fiber Delay Line) is proposed. Firstly, the incident radiofrequency signal is modulated on light wave, which is then divided into several optical paths by optical power divider. These light waves are transmitted in optical fibers with different length, thus creating the required delays of electrical signals after detected by the photodetectors. The detected electrical signals channels are subsequently connected to a radio anti-saturation module which is based on the multiple radio channels and the LMS algorithm. The simulation results demonstrate that the power of interference can be suppressed by about 60 dB, indicating that the effect of anti-saturation is remarkable, and the single-channel saturation problem of receivers is then solved effectively.

In this paper we propose a Least Mean Square (LMS) algorithm for the practical training of the class of min-max classifiers. These are lattice-theoretic generalization of Boolean functions and are also related to feed-forward neural networks and morphological signal operators. We applied the LMS algorithm to the problem of handwritten character recognition. The database consists of segmented and cleaned digits. Features that were extracted from the digits include Fourier descriptors and morphological shape-size histograms. Experimental results using the LMS algorithm for handwritten character recognition are promising. In our initial experimentation, we applied the min-max classifier to binary classification of '0' and '1' digits. By preprocessing the feature vectors, we were able to achieve an error rate of 1.75% for a training set of size 1200 (600 of each digit); and an error rate of 4.5% on a test set of size 400 (200 of each). These figures are comparable to those obtained by 2-layer neural nets trained using back propagation. The major advantage of min-max classifiers compared to neural networks is their simplicity and the faster convergence of their training algorithm.