Least Mean Square Error Research Articles

In-Body to On-Body Channel Characterization and Modeling Based on Heterogeneous Human Models at HBC-UWB Band

In this work, the in-body to on-body (IB2OB) channel characteristics of the ultrawideband (UWB) link from 10 to 50 MHz at the human body communication (HBC) band are studied for developing high-speed and large-capacity intrabody wireless communication systems. Incorporating the time-domain finite integration technique (FIT) with three types of anatomical human models (adult male, adult female, and teenager) and a simple geometric human model with homogeneous or multilayer biological tissues, the 10–50-MHz signal propagation characteristics are investigated, and a comprehensive path loss model is proposed based on typical in-body links of the heart, brain, and intestine to the on-body sensor nodes. The statistical results show that a linear regulation term related to the surface wave propagation should be added to construct a more accurate logarithmic-distance path loss model in the sense of least-mean-square error (LMSE), indicating that the surface wave propagation mechanism can describe the 10–50-MHz signal propagation more accurately. The shadow fading in decibel is normally distributed around the mean path loss with a standard deviation in the range of 7–9 dB. And the average group delay of the IB2OB channel fluctuates around 0–7 ns, illustrating that the channel can be approximated as a linear phase response. Moreover, the experimental verification is conducted with a tissue-equivalent liquid phantom and an adult man, demonstrating the validity of our proposed statistical path loss model. The key parameters for channel characterization in simulations and measurements are summarized to provide useful insights for HBC-UWB communication system design and performance analysis.

An Adaptive Real-Time Technique for Harmonics Estimation Using Adaptive Radial Basis Function Neural Network

In this paper, a neural networks algorithm based on adaptive radial basis function (ARBF) is used to decompose the grid current drawn by nonlinear load, and the fundamental and harmonic components are estimated. The learning rate – considered as one of the most important parameters that govern the performance of the ARBF network - is investigated as well to reduce the system total error. Two methodologies are proposed to improve the estimation of the fundamental component of highly nonlinear current signal. One is based on fast Fourier transform (FFT) and the other is based on least mean square error (LMSE). The error between the reference signal and the reproduced signal (the sum of estimated fundamental and harmonic signals) is chosen as performance index. The obtained results unveil that both methodologies can be effective in enhancing the system accuracy, and that the proposed algorithm can provide better performance compared to the conventional RBF network.

An Ultra-short-term Wind Power Forecasting Method with Special Error Distribution

The randomness and uncontrollability of wind energy have brought many difficulties to the development of wind power generation. How to obtain accurate forecasting results of wind power has become an increasingly important topic these years. This paper proposes an ultra-short-term wind power forecasting method using Elman neural network with an enhanced gradient descent training algorithm, which is able to compute the probabilistic distribution function (PDF) of the forecasting error by utilizing two kinds of series expansion for the purpose of regarding it as the loss function of the forecasting model. To validate the accuracy and efficiency of the purposed method, the conventional least mean square error (LMSE) based model is served as a benchmark in the comparison of simulation results. At last, historical wind power statistics of one month collected from a wind farm in the northern Hebei Province of China are used to perform single-point and probabilistic predictions in order to verify the effectiveness of the purposed method.

Epileptic seizure detection using novel Multilayer LSTM Discriminant Network and dynamic mode Koopman decomposition

Epilepsy is a neurological disorder that causes abnormality in brain function and leads to unusual behavior, awareness loss and other defects. An Electroencephalogram (EEG) is an electrophysiological screening method used for diagnosing epileptic seizure and further brain activities. Usually, physicians inspect the brain abnormalities. But this technique is highly time-consuming, has poor consistency, and has faced difficulty in detecting the seizure due to the imbalanced distribution of data. To overcome these issues a novel seizure predicting algorithm has been proposed. Initially, data is pre-processed using modified common spatial pattern (CSP) which will convert various channels into alternate channels. This in turn results in a maximized signal to noise ratio. For feature extraction, Dynamic Mode Koopman Decomposition (DMKD) and Empirical Mode Decomposition (EMD) has been utilized by obtaining multiple IMFs (Intrinsic Mode Functions). A scheme called Novel Multilayer LSTM Discriminant Network is used for the classification process and is used for efficiently classifying featured vectors. The classification process is used for classifying ictal and non-ictal regions. To avoid misinterpretations and to reduce false alarm rate, post-processing is done. The performance of the proposed system is analyzed using various parameters such as accuracy, specification, sensitivity, Jaccard and other parameters. The proposed is compared with existing methods namely LMS (Least Mean Square) and Wiener in terms of PSNR (Peak Signal-to-Noise Ratio), LMSE (Least Mean Square Error), Contrast-to-Noise Ratio (CNR) and error rates. From performance analysis, it is clear that the proposed framework with an efficient algorithm will provide effective detection for seizure and non-seizure patients.

Machinability and cutting force modeling of 7055 aluminum alloy with wide temperature range based on dry cutting

The machinability of 7055 aluminum alloy with wide temperature range is examined, with focus on the three cutting forces, surface quality and work hardening of the material under low, medium, and high temperatures. The results demonstrate that, under low temperature, the work hardening depth of 7055 aluminum alloy is almost insensitive to the cutting speed, whereas at a higher cutting speed, the work hardening degree of the material first decreases and then increases; both the work hardening degree and hardening depth are significantly positively correlative to the cutting depth: the work hardening degree is positively correlative, though not so significantly, to the feed rate, while the work hardening depth is insensitive to the feed rate and remains at 100 μm in all cases. Under high temperature, the work hardening degree of 7055 aluminum alloy is positively correlative to the cutting speed; at depths smaller than 80 μm below the machined surface, the work hardening degree is negatively correlative to the cutting depth; at depths larger than 80 μm below the cutting surface, the work hardening degree of the material becomes significantly positively correlative to the cutting depth. A mathematical model of three cutting forces in dry cutting with wide temperature range is established based on wide temperature-range dynamic impact experimental results and the orthogonal cutting model, and modified using the LMSE (least mean square error) principle. The errors between the predicted and experimental three cutting forces, after modification, are all smaller than 10%, which is within the permissible limit of error. This verifies that the modified three cutting force prediction model can predict cutting forces accurately.

A New Recognition and Pose Estimation of Tiny Electronic Parts Using Principal Component Analysis and Harris Corner Features

The center coordinates and rotated angle of the electronic part must be precisely estimated in case of palletizing the parts by picking and placing the parts to output tray by a robot, especially, the tiny parts like surface mounting device (SMD). A machine vision algorithm must be used in recognizing the type as well as estimating the pose of the parts. In this paper, we propose not only a recognition algorithm of electronic part using Principal Component Analysis [PCA] combined with Harris corner feature (HCF) and least mean square error (LMSE) for the matching of the part with data base, but also a precise pose estimation of the recognized part by using the extracted four corners from HCF combined with lines from Hough transformation (LHT). Applying the proposed algorithm to the test setup of five kinds of SMDs, we verified the usefulness of the proposed algorithm in accurate pose estimation as well as in recognition of type of the parts even in environment with illumination changes.

Multiclass Alpha Integration of Scores from Multiple Classifiers.

Alpha integration methods have been used for integrating stochastic models and fusion in the context of detection (binary classification). Our work proposes separated score integration (SSI), a new method based on alpha integration to perform soft fusion of scores in multiclass classification problems, one of the most common problems in automatic classification. Theoretical derivation is presented to optimize the parameters of this method to achieve the least mean squared error (LMSE) or the minimum probability of error (MPE). The proposed alpha integration method was tested on several sets of simulated and real data. The first set of experiments used synthetic data to replicate a problem of automatic detection and classification of three types of ultrasonic pulses buried in noise (four-class classification). The second set of experiments analyzed two databases (one publicly available and one private) of real polysomnographic records from subjects with sleep disorders. These records were automatically staged in wake, rapid eye movement (REM) sleep, and non-REM sleep (three-class classification). Finally, the third set of experiments was performed on a publicly available database of single-channel real electroencephalographic data that included epileptic patients and healthy controls in five conditions (five-class classification). In all cases, alpha integration performed better than the considered single classifiers and classical fusion techniques.

Adaptive Genetic Algorithm Based Multi-Objective Optimization for Photovoltaic Cell Design Parameter Extraction

In this paper, an enhanced evolutionary computing algorithm has been attempted for photo voltaic (PV) design parameter extraction using adaptive genetic algorithm. The I-V curve fitting approach has been used to find optimal photovoltaic parameters Unlike single objective function based approaches, multiple objective functions including, least mean square error and Pearson residual error optimization are considered to fit the I-V curve. A cumulative fitness function is derived using both objectives that alleviate computational complexity, local minima and convergence. Importantly, Pearson residual error optimization (PRO) optimizes least mean square error (LSE) reduction while alleviating the probability of under/over-fitting that ensures optimal PV design parameter identification

A Novel Strategy for Proficient Channel Limitation Equalizer

A time-domain equalizer (TEQ) employed in cascade with the channel yields an actual impulse response that is squatter than the channel impulse response. Time-domain equalization is critical in reducing channel state dimension in maximum likelihood sequence approximation, and inter-carrier and inter-symbol interference in multicarrier systems. This project evaluates two TEQ design methods docile to cost-effective real-time implementation: least mean squared error (LMSE) and extreme shortening SNR (ESSNR) methods. We condense the complexity of computing the matrices in the ESSNR and LMSE designs by a factor of 140 and a factor of 16 (respectively) relative to existing approaches, without degrading performance. We prove that an infinite length ESSNR TEQ with unit norm TEQ constraint is symmetric. A symmetric TEQ halves FIR implementation density, enables parallel training of the frequency-domain equalizer and TEQ, reduces TEQ training complexity by a factor of 4 and doubles the length of the TEQ that can be designed using fixed-point arithmetic, with only a small loss in bit rate. Simulations are presented for designs with a symmetric TEQ or target impulse response.

A NON-ITERATIVE METHOD FOR FACTORIZATION OF POSITIVE MATRIX IN DISCRETE WAVELET TRANSFORM BASED IMAGE COMPRESSION

A non-iterative method of factorizing a 4◊4 positiv e matrix, with the application to image compression is explained using an example. The procedure is applie d to all the 4096 number of 4◊4 pixel sub-blocks of a 256◊256 image for compression. The proposed compression technique can be applied to the Discrete Wavelet Transform (DWT) coefficients of the test image. The 16 Pixel Intensity Values (PIV) or their D WT coefficients of a 4◊4 pixels sub-block of the image can be represented by the outer product of a 4◊1 c olumn matrix and a 1◊4 row matrix, with Least Mean Square Error (LMSE) criterion. Hence, instead of transmitting the 16 PIVs or their DWT coefficients, the values of the 4 elements of the column matrix and the 4 elements of the row matrix alone are transmit ted resulting in a maximum compression ratio of 2 (16/4+4). The receiver can recreate the 4◊4 pixels sub-block or their DWT coefficients, by calculating the outer products of 4 values of column matrix with 4 values of row matrix. In case of DWT coefficients inverse DWT is applied to recreate the pixels. This principle is extended to all the sub-blocks of the 256◊256 image to compress and later reconstruct the image.

Optimizing Outdoor Propagation Model based on Measurements for Multiple RF Cell

This paper intends to study outdoor RF attenuation path loss behavior under certain restrictions. The study has been conducted in Nablus city to develop and optimize a suitable propagation model based on one of the existing propagation models based on outdoor measurements for 900 MHz, where a local GSM system is operating under sever geographical terrains and frequency limitations. The optimized model has been chosen such that certain error parameters are minimized. Some of the proposed models are; Bertoni-Walfish, Hat, Walfisch-Ikegami and the standard macrocell. In this paper a Tuned Bertoni-Walfisch model has outperformed the other models and has proven, to be the best suited for propagation analysis involving such terrain. This is achieved by varying the range dependence using Least Mean Square Error (LMSE) method

Optimal algorithm for FIR digital filter with canonical signed digit coefficients

In order to save the resources of the Finite Impulse Response(FIR) filter and increase the running speed,it was proposed to use the Least Mean-Square-Error(LMSE) to transfer the float point coefficients filter to the Canonical Signed Digit(CSD) filter.The FIR filter was implemented by the cascades structure,which conjugated pairs of zeros into two basic sections.First,all zeros of the digital filter were calculated,which were made of two cascade sections for an FIR.And then the coefficients of the first cascade were transferred to fixed point.Next step was to quantize the second cascade coefficients into fixed point.To eliminate the finite word-length effects,the LMSE was adopted to compensate zeros in this step.Finally,all the fixed point coefficients were quantized into CSD.In order to prove the effectiveness of the two methods,and the FIR filter was also designed with simple quantized coefficients.The magnitude responses of two methods show that the LMSE quantization is more effective than that of the simple quantization.

A Class of Adaptive Algorithms Based on Entropy Estimation Achieving CRLB for Linear Non-Gaussian Filtering

Adaptive filtering has been extensively studied under the assumption that the noise is Gaussian. The most commonly used least-mean-square-error (LMSE) filter is optimal when the noise is Gaussian. However, in many practical applications, the noise can be modeled more accurately using a non-Gaussian distribution. In this correspondence, we consider non-Gaussian distributions for the noise model and show that the filter of using entropy bound minimization (EBM) leads to significant performance gain compared to the LMSE filter. The least mean p-norm (LMP) filter using the α-stable distribution to model noise is shown to be the maximum-likelihood solution when using the generalized Gaussian distribution (GGD) to model noise. The GGD model for noise allows us to compute the Cramer-Rao lower bound (CRLB) for the error in estimating the weights. Simulations show that both the EBM and LMP filters achieve the CRLB as the sample size increases. The EBM filter is shown to be less committed with respect to unseen data yielding generally superior performance in online learning when compared to LMP. We also show that, when the noise comes from impulsive α -stable distributions, both the EBM and LMP filters provide better performance than LMSE. In addition, the EBM filter offers the advantage that it does not assume a certain parametric model for the noise, and by proper selection of the measuring functions, it can be adapted to a wide range of noise distributions.

A Study on Fusion of Different Resolution Images

How to fuse different resolution images of the same target obtained by multiple sensors is very important in the field of image fusion. In this paper, a method of fusing different resolution images is proposed based on least mean square error (LMSE).The main idea of the algorithm is to transform the image fusion problem into a constrained optimization problem. Experimental results show that the proposed method exhibits a good performance.

Reliability Analysis for Global Motion Estimation

Global motion estimation (GME) is the enabling step for many important video exploitation tasks. In this work, we focus on indirect GME methods which have low computational complexity. Typically, an indirect GME method has two major steps. The first step is to find point correspondence between frames through local motion search or feature matching. Then, the second step determines global motion parameters using optimal model fitting, such as least mean-squared error (LMSE) fitting or RANSAC. However, due to image noise and inherent ambiguity in point correspondence, local motion estimation often suffers from relatively large errors, which degrade the performance and reliability of GME. In this work, we propose a method to characterize the reliability of local motion estimation results and use this reliability measure as a weighting factor to determine the importance level of each local motion estimation result during global motion estimation. Our simulation results demonstrate that the proposed scheme is able to significantly improve the accuracy and robustness of global motion estimation with a very small computational overhead.

LMSE-based parameter acquisition for multicarrier CDMA systems

An acquisition scheme, based on the least mean-squared error (LMSE), for the symbol timing, carrier phases, and multipath gains for multicarrier code-division multiple-access (MC-CDMA) systems is proposed. A new coefficient called the quasi-mean-squared accuracy coefficient (QMSAC) is introduced. This allows the search for multiple optimal parameters to be split into simpler single-parameter estimates. The symbol timing is acquired with a comparison of the QMSACs followed by a linear interpolation scheme or a search of a minimizing variable. Carrier phases and multipath gain coefficients are then estimated via some algebraic computations based only on the estimated symbol timing and the QMSACs. The spreading sequence of the desired user is the only prior knowledge required for the proposed approach. Simulation results demonstrate the near-far resistance of the approach and show that the error performance of minimum mean-squared error (MMSE) detection with acquired parameters is only a few decibels worse for the considered channel than the performance obtained with perfect knowledge about the channel impulse response.

Joint time difference of arrival/angle of arrival position finding in passive radar

Several target position finding methods are proposed in various papers mainly regarding sensor networks. However, the problem of position finding in passive radar systems is somewhat different from the general case of sensor networks. Generally, in a passive radar system, there are few receivers located at short distances when compared with the it distance from the target. In this case, a problem known as geometric dilution of precision (GDP) occurs, which considerably increases the error of many proposed methods. This phenomenon can even make the position finding equations non-solvable. Regarding this fact, we have developed a least mean square error (LMSE) based method, which uses both the angle of arrival (AOA) and time difference of arrival (TDOA) measurements to estimate almost precisely the position of very distant targets in a passive radar system. A simple equation is derived here to compare the TDOA and AOA accuracies. Then, it is shown that whenever the accuracies of these two measurements are comparable, the proposed method estimates the target position more precisely than the conventional AOA-only and TDOA-only methods. It also avoids the non-convergence behaviour encountered in TDOA-only methods.

Algorithms for active vibration isolation on spacecraft using a Stewart platform

This paper presents the recent development in algorithms for active vibration isolation onspacecraft using a Stewart platform. The multiple error least mean square (LMS) algorithmand the clear box algorithm have been implemented on these platforms and severalenhancements have been made to the clear box algorithm. Based on experimental results, itis concluded that the multiple error LMS algorithm is preferred for vibration isolation whena disturbance correlated signal is available. In the absence of such a signal, the clear boxalgorithm is the method of choice. Among the implementations of the clear boxalgorithm, the sine/cosine method is preferred for handling time-invariant disturbancefrequencies, the adaptive method for rapidly varying disturbance frequencies, andthe frequency-domain method for a large number of time invariant disturbancefrequencies.

Weighted DPOAE Input/Output-functions: A tool for automatic assessment of hearing loss in clinical application

2f1 – f2 distortion product otoacoustic emission (DPOAE) input/output functions were recorded in 796 ears with sensorineural hearing loss at up to 50 frequencies between 500 Hz and 8 kHz in a wide level range from L2 = 20 to L2 = 65 dB SPL at up to 10 levels (“scissor paradigm”, L1 = 0.4L2 + 39, f2/f1 = 1.2). DPOAEs were accepted as valid for signal-to-noise ratios (SNR) only if they exceeded 6 dB. To assess the DPOAE threshold, DPOAE input/output functions (L2 – DPOAE sound pressure diagram) were subject to linear regression and extrapolation; the point of intersection with the L2 coordinate was then interpreted as DPOAE threshold [1]. The linear fit of the function pDP (L2) = a + bL2 (whereby a and b represent the threshold and the slope of the DPOAE growth, respectively), reflects compression and sensitivity of the cochlear amplifier. Using a weighted least mean square error (LMSE) regression, more than 70 % of the DPOAE I/O functions of the data set can be used for the threshold estimation. High correlation and relatively small differences between objectively estimated DPOAE threshold and subjectively determined hearing threshold of pure-tone audiograms prove that the weighted extrapolation of DPOAE I/O functions can be a valuable clinical tool for the objective assessment of cochlear hearing loss.

FINDING UNDERLYING FACTORS IN TIMESERIES

We compare four neural methods of pre-processing time series data - Principal Component Analysis (PCA) (Karhunen and Joutsensalo 1994), a neural implementation of Factor Analysis (FA), Independent Component Analysis (ICA) (Hyva¨rinen and Oja 1997) and Complexity Pursuit (CP) (Hyva¨rinen 2001) - with a view to subsequently using a multi-layer perceptron (MLP) to forecast on the data set. Our rationale is that forecasting the underlying factors will be easier than forecasting the original time series which is a combination of these factors. The projections of the data onto the filters found by the pre-processing method were fed into the MLP and it was trained to find Least Mean Square Error (LMSE). We show that forecasting the projections on the underlying factors reduces the need to consider the possibility of overtraining the MLP. The last method, CP, achieves by far the best (in terms of least mean square error) performance. Factor Analysis (FA) and particularly Independent Component Analysis (ICA) have the worst performance. Minor modifications to the Complexity Pursuit (CP) method are shown to improve performance in terms of LMSE.