Estimation algorithm for system with non‐Gaussian multiplicative/additive noises based on variational Bayesian inference

This paper considers estimation algorithms for linear and nonlinear systems contaminated by non‐Gaussian multiplicative and additive noises. Based on the variational idea, in order to derive optimal estimation algorithms, we combine the multiplicative noise with states as the joint parameters to estimate. The application of variational Bayesian inference to joint estimation of the state and the multiplicative noise is established. By treating the states as unknown quantities as well as the multiplicative noise, there are now correlations between the states and multiplicative noise in the posterior distribution. There are two main goals in Bayesian learning. The first is approximating the marginal likelihood (PDF of multiplicative noise) to perform model comparison. The second is approximating the posterior distribution over the states (also called a system model), which can then be used for prediction. The two goals constitute the iterative algorithm. The rules for determining the loop is the Kullback‐Leibler divergence between the true distribution of state and a chosen fixed tractable distribution, which is used to approximate the true one. The iterative algorithm is deduced, which is initialized based on the idea of sampling. Meanwhile, the convergence analysis of the proposed iterative algorithm is presented. The numerical simulation results in a comparison between the proposed method and these existing classic algorithms in the context of nonlinear hidden Markov models, state‐space models, and target‐tracking models with non‐Gaussian multiplicative noise demonstrate the superiorities, not only in speed, precision, and computation load but also in the ability to process non‐Gaussian complex noise.     

A new algorithm for shot noise removal in medical ultrasound images based on alpha‐stable model

Ultrasonic images are generally affected by multiplicative shot noise. Shot noise filtering is thus a critical pre‐processing step in medical ultrasound imagery. This paper analyses and models the coefficients of 2‐D multi‐resolution wavelet decomposition of logarithmically transformed images using alpha‐stable distribution model. Consequently, we propose a new function that performs a non‐linear operation on the data of classifying the coefficients, thus achieving a novel form of noise removal based on multi‐resolution wavelet decomposition and the alpha‐stable model. We compare our new technique with current shot noise reduction methods applied on actual ultrasound medical images and simulations results show that the proposed new method is more robust than the methods based on Gaussian assumption. Copyright © 2006 John Wiley & Sons, Ltd.     

一种视频图像序列人脸检测方法

根据高阶统计量对高斯噪声不敏感的特点,本文提出了一种新的从连续图像序列中检测人脸的方法.首先,设计了运动图像序列帧间差的高阶统计量的计算公式,以自动分离运动人体和背景.为了提高速度,对计算过程进行了优化处理.然后使用一种边缘人脸检测方法检测运动人体中的人脸,并对该人脸进行尺度归一化处理.实验表明,该方法简单易行,速度快,能准确地提取出运动图像序列中的人脸.     

Exploiting all combinations of microphone sensors in overdetermined frequency domain blind separation of speech signals

A new approach to overdetermined frequency domain blind source separation (BSS) of speech signals which exploits all combinations of observations and hence varying inter microphone spacings is proposed. The observations are divided into subgroups so that conventional frequency domain BSS algorithms can be used. By evaluating the separation performance obtained from each group on the basis of approximately measuring the independence of separated signals, the output of the group that has the best performance among all groups on a frequency‐by‐frequency basis is chosen as the overall output. The separated signals of the overall system are then obtained by transforming their frequency domain representations into the time domain. Simulation results based on speech signals confirm that the proposed approach has better performance based on the performance index (PI) as compared with a conventional scheme using only one microphone group and an existing overdetermined frequency domain BSS algorithm. Copyright © 2010 John Wiley & Sons, Ltd.     

Second‐order intermodulation cancelation and conversion‐gain enhancement techniques for CMOS active mixers

In this paper, two new techniques are proposed to improve the second‐order input intercept point (IIP2) and conversion‐gain in double‐balanced Gilbert‐cell complementary metal‐oxide semiconductor (CMOS) mixers. The proposed IIP2 improvement technique is based on canceling the common‐mode second‐order intermodulation (IM2) component at the output current of the transconductance stage. Additionally, the conversion‐gain is improved by increasing the fundamental component of the transconductance stage output current and creating a negative capacitance to cancel the parasitic capacitors. Moreover, in the proposed IM2 cancelation technique, by decreasing the bias current of the switching transistors, the flicker noise of the mixer is reduced. The proposed mixer has been designed with input frequency and output bandwidth equal to 2.4 GHz and 20 MHz, respectively. Spectre‐RF simulation results show that the proposed techniques simultaneously improve IIP2 and conversion‐gain by approximately 23.2 and 5.7 dB, respectively, in comparison with the conventional mixer with the same power consumption. Also, the noise figure (NF) at 20 kHz, where the flicker noise is dominant, is reduced by 4.9 dB. The average NF is increased nearly 0.9 dB, and the value of third‐order input intercept point (IIP3) is decreased approximately 1.8 dB. Copyright © 2014 John Wiley & Sons, Ltd.     

改进的带二阶项配电网快速潮流算法

在简要分析已有配电网潮流算法的基础上,提出了一种改进的带二阶项的快速潮流算法.该算法应用矩阵分块求逆方法对阶数较高的雅可比阵求逆计算进行了改进,使阶数较高的雅可比阵的求逆变为阶数较低的四个阵的求逆.在充分发挥带二阶项的快速潮流算法收敛性好的基础上,提高了计算速度,适用于各种复杂的配电网络,并基于该算法开发了配电网潮流计算软件.27节点和33节点算例计算结果表明,该方法具有良好的收敛性,能很好地处理配电网重构,具有较高的计算速度.     

Bias‐eliminating least‐squares identification of errors‐in‐variables models with mutually correlated noises

This paper proposes a bias‐eliminating least‐squares (BELS) approach for identifying linear dynamic errors‐in‐variables (EIV) models whose input and output are corrupted by additive white noise. The method is based on an iterative procedure involving, at each step, the estimation of both the system parameters and the noise variances. The proposed identification algorithm differs from previous BELS algorithms in two aspects. First, the input and output noises are allowed to be mutually correlated, and second, the estimation of the noise covariances is obtained by exploiting the statistical properties of the equation error of the EIV model. Copyright © 2012 John Wiley & Sons, Ltd.     

Robust intelligent PCA‐based face recognition framework using GNP‐fuzzy data mining

Traditional principal component analysis (PCA) based face recognition algorithms have a low recognition accuracy due to the influence of noise and illumination changes. This paper proposes a robust, intelligent PCA‐based face recognition framework in the complicated illumination database when using multiple training images per person (MTIP‐CID). There are mainly two improvements in the proposed method. One is that a face‐recognition‐oriented genetic‐based clustering algorithm is introduced to reduce the influence of a large number of classes on the classification accuracy in the MTIP‐CID. The other is that a classifier based on fuzzy class association rules (FCARs) is applied to mine the inherent relationships between eigenfaces and to improve the robustness of PCA‐based face recognition in noisy environments. Experimental results on the extended Yale‐B database demonstrate that the proposed framework performs better and is more robust against noise compared with other traditional face recognition algorithms, i.e. linear discriminant analysis (LDA) and local binary patterns (LBPs). © 2013 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.     

Convergence properties of bias‐eliminating algorithms for errors‐in‐variables identification

This paper considers the problem of dynamic errors‐in‐variables identification. Convergence properties of the previously proposed bias‐eliminating algorithms are investigated. An error dynamic equation for the bias‐eliminating parameter estimates is derived. It is shown that the convergence of the bias‐eliminating algorithms is basically determined by the eigenvalue of largest magnitude of a system matrix in the estimation error dynamic equation. When this system matrix has all its eigenvalues well inside the unit circle, the bias‐eliminating algorithms can converge fast. In order to avoid possible divergence of the iteration‐type bias‐eliminating algorithms in the case of high noise, the bias‐eliminating problem is re‐formulated as a minimization problem associated with a concentrated loss function. A variable projection algorithm is proposed to efficiently solve the resulting minimization problem. A numerical simulation study is conducted to demonstrate the theoretical analysis. Copyright © 2005 John Wiley & Sons, Ltd.     

Signal separation using second‐ and high‐order statistics

This paper presents two methods for signal separation. In either method, the fundamental criterion for separation relies on reducing to zero, or at least minimizing, the output cross‐correlation or cross‐cumulant functions of a decoupling multi‐input–multi‐output system that is fed with mixed signals. In one of the approaches used, the parameters of this system are determined through solving — in a least‐squares sense — a linearized set of equations describing the deviations from zero of either the cross‐correlation or cross‐cumulant functions when evaluated for different lags. An alternative rapidly convergent adaptive algorithm is also described for minimizing the cross‐correlation or cross‐cumulant functions. The paper also considers both FIR and IIR representations of the decoupling system. It shows that using IIR functions in the decoupling system does not offer any merit over the FIR case. Illustrative examples are given to show the performance of the proposed algorithms. Copyright © 2000 John Wiley & Sons, Ltd.     

Second‐order blind algorithms for adaptive recursive linear equalizers: analysis and lattice implementation

Adaptive recursive linear equalizers present important advantages in terms of performance and robustness compared to more standard finite impulse response structures, and provide a means for blindly initializing the decision feedback structure. We present an analysis of a pair of algorithms for the adaptation of the recursive part of the equalizer, which are based on the second‐order statistics of the received signal, in a multichannel complex‐valued setting with spatially coloured noise. When the number of equalizer poles is no less than the channel order, both algorithms enjoy a unique stationary point, which in addition is locally convergent; global convergence properties, on the other hand, can be quite different. When the optimum setting presents poles close to the stability boundary, the lattice structure is preferred for ease of stability monitoring. Lattice versions of the two algorithms are developed and their convergence properties discussed. Copyright © 2006 John Wiley & Sons, Ltd.     

Convergence of the recursive identification algorithms for multivariate pseudo‐linear regressive systems

The performance analysis of the recursive algorithms for the multivariate systems with an autoregressive moving average noise process is still open. This paper analyzes the convergence of two recursive identification algorithms, the multivariate recursive generalized extended least squares algorithm and the multivariate generalized extended stochastic gradient algorithm, for pseudo‐linear multivariate systems and proves that the parameter estimation errors consistently converge to zero under persistent excitation conditions. The simulation results show that the proposed algorithms work well. Copyright © 2015 John Wiley & Sons, Ltd.     

Second order statistics of the SC receiver over Rician fading channels in the presence of multiple Nakagami‐m interferers

In this paper, a numerical approach to the performance analysis of a signal‐to‐interference‐based selection combining receiver, operating over a Rician multipath‐fading environment, in the presence of multiple Nakagami‐m co‐channel interferers is provided. Closed form expressions are obtained for the first‐order and second‐order statistical measures, that is, the probability density function, cumulative distribution function, and average level‐crossing rate of the received signal. Further, the aforementioned results are used for an efficient evaluation of other performance measures, such are outage probability and average fade duration. A graphical representation of the obtained numerical results is also provided in order to present and discuss the influence of transmission parameters on the studied performance measures. Copyright © 2015 John Wiley & Sons, Ltd.     

A wideband time‐based continuous‐time sigma‐delta modulator with 2nd order noise‐coupling based on passive elements

Embedding the time encoding approach inside the loop of the sigma‐delta modulators has been shown as a promising alternative to overcome the resolution problems of analog‐to‐digital converters in low‐voltage complementary metal‐oxide semiconductor (CMOS) circuits. In this paper, a wideband noise‐transfer‐function (NTF)‐enhanced time‐based continuous‐time sigma‐delta modulator (TCSDM) with a second‐order noise‐coupling is presented. The proposed structure benefits from the combination of an asynchronous pulse width modulator as the voltage‐to‐time converter and a time‐to‐digital converter as the sampler to realize the time quantization. By using a novel implementation of the analog‐based noise‐coupling technique, the modulator's noise‐shaping order is improved by two. The concept is elaborated for an NTF‐enhanced second‐order TCSDM, and the comparative analytical calculations and behavioral simulation results are presented to verify the performance of the proposed structure. To further confirm the effectiveness of the presented structure, the circuit‐level implementation of the modulator is provided in Taiwan Semiconductor Manufacturing Company (TSMC) 90 nm CMOS technology. The simulation results show that the proposed modulator achieves a dynamic range of 84 dB over 30 MHz bandwidth while consuming less than 25 mW power from a single 1 V power supply. With the proposed time‐based noise‐coupling structure, both the order and bandwidth requirements of the loop filter are relaxed, and as a result, the analog complexity of the modulator is significantly reduced. Copyright © 2015 John Wiley & Sons, Ltd.     

Real‐valued MUSIC algorithm for power harmonics and interharmonics estimation

Recently, advanced spectrum estimation methods, including the MUSIC (Multiple Signal Classification) algorithms, are being gradually employed for high‐resolution power harmonics analysis. However, most of them are proposed to detect frequencies of complex‐valued signals, so that any real‐valued signal should be transformed into complex form. This data pre‐treatment may lead to additional computation burden. In addition, the picket‐fence effects also exist as in the FFT algorithm and cause poor frequency resolution. To overcome these drawbacks, a real‐valued MUSIC algorithm is proposed for power harmonics analysis in this paper. The algorithm is based on the subspace decomposition theory and the computation of pseudospectrum is also provided. Additionally, to improve the measuring precision, the Newton–Raphson algorithm is adopted to optimize the harmonic frequencies significantly. Simulation results show that, in the real‐valued MUSIC pseudospectrum, the spectral peaks of actual harmonic components can be more easily distinguished from the false peaks caused by noise, and the computational complexity is notably lower than that of the classic complex MUSIC, as well as the detecting accuracy is close to that of root‐MUSIC algorithm which is quite time consuming. Experimental results prove that the proposed strategy is more suitable for high‐resolution power harmonics estimation. Copyright © 2010 John Wiley & Sons, Ltd.     

一种基于信息最大化的自适应变步长盲源分离算法

现有的盲源分离算法不适合于数据的实时处理,并且算法性能依赖于步长的选择.提出一种基于信息最大化的自适应变步长盲源分离算法,采用基于估计函数的变步长算法,降低了盲源分离算法性能对步长的依赖性,并且采用自适应处理形式,适合数据的实时处理.最后将其应用于声音信号的盲分离,在选择小的步长参数的情况下,原有算法和文中新算法都取得了良好的分离效果;在选择较大的步长参数的情况下,新算法优于传统算法.     

Finite difference schemes for heat conduction analysis in integrated circuit design and manufacturing

The importance of thermal effects on the reliability and performance of VLSI circuits has grown in recent years. The heat conduction problem is commonly described as a second‐order partial differential equation (PDE), and several numerical methods, including simple explicit, simple implicit and Crank–Nicolson methods, all having at most second‐order spatial accuracy, have been applied to solve the problem. This paper reviews these methods and further proposes a fourth‐order spatial‐accurate finite difference scheme to better approximate the PDE solution. Moreover, we devise a fourth‐order accurate approximation of the convection boundary condition, and apply it to the proposed finite difference scheme. We use a block cyclic reduction and a recently developed numerically stable algorithm for inversion of block‐tridiagonal and banded matrices to solve the PDE‐based system efficiently. Despite their higher computation complexity than direct computation in a sequential processor, we make it possible for the very first time to employ a divide‐and‐conquer algorithm, viable for parallel computation, in heat conduction analysis. Experimental results prove such possibility, suggesting that applying divide‐and‐conquer algorithms, higher‐order finite difference schemes can achieve better simulation accuracy with even faster speed and less memory requirement than conventional methods. Copyright © 2010 John Wiley & Sons, Ltd.     

An algorithm for direct identification of passive transfer matrices with positive real fractions via convex programming

The paper presents a new algorithm for the identification of a positive real rational transfer matrix of a multi‐input–multi‐output system from frequency domain data samples. It is based on the combination of least‐squares pole identification by the Vector Fitting algorithm and residue identification based on frequency‐independent passivity constraints by convex programming. Such an approach enables the identification of a priori guaranteed passive lumped models, so avoids the passivity check and subsequent (perturbative) passivity enforcement as required by most of the other available algorithms. As a case study, the algorithm is successfully applied to the macro‐modeling of a twisted cable pair, and the results compared with a passive identification performed with an algorithm based on quadratic programming (QPpassive), highlighting the advantages of the proposed formulation. Copyright © 2010 John Wiley & Sons, Ltd.     

Development of numerical linear algebra algorithms in dynamic fixed‐point format: a case study of Lanczos tridiagonalization

Proper range and precision analysis play an important role in the development of fixed‐point algorithms for embedded system applications. Numerical linear algebra algorithms used to find singular value decomposition of symmetric matrices are suitable for signal and image‐processing applications. These algorithms have not been attempted much in fixed‐point arithmetic. The reason is wide dynamic range of data and vulnerability of the algorithms to round‐off errors. For any real‐time application, the range of the input matrix may change frequently. This poses difficulty for constant and variable fixed‐point formats to decide on integer wordlengths during float‐to‐fixed conversion process because these formats involve determination of integer wordlengths before the compilation of the program. Thus, these formats may not guarantee to avoid overflow for all ranges of input matrices. To circumvent this problem, a novel dynamic fixed‐point format has been proposed to compute integer wordlengths adaptively during runtime. Lanczos algorithm with partial orthogonalization, which is a tridiagonalization step in computation of singular value decomposition of symmetric matrices, has been taken up as a case study. The fixed‐point Lanczos algorithm is tested for matrices with different dimensions and condition numbers along with image covariance matrix. The accuracy of fixed‐point Lanczos algorithm in three different formats has been compared on the basis of signal‐to‐quantization‐noise‐ratio, number of accurate fractional bits, orthogonality and factorization errors. Results show that dynamic fixed‐point format either outperforms or performs on par with constant and variable formats. Determination of fractional wordlengths requires minimization of hardware cost subject to accuracy constraint. In this context, we propose an analytical framework for deriving mean‐square‐error or quantization noise power among Lanczos vectors, which can serve as an accuracy constraint for wordlength optimization. Error is found to propagate through different arithmetic operations and finally accumulate in the last Lanczos vector. It is observed that variable and dynamic fixed‐point formats produce vectors with lesser round‐off error than constant format. All the three fixed‐point formats of Lanczos algorithm have been synthesized on Virtex 7 field‐programmable gate array using Vivado high‐level synthesis design tool. A comparative study of resource usage and power consumption is carried out. Copyright © 2015 John Wiley & Sons, Ltd.     

State estimation of discrete‐time systems with arbitrarily correlated noises

In this paper, the state estimation problem for discrete‐time systems is considered where the noises affecting such systems do not require any constraint condition for the correlation and distribution, that is, the noises can be arbitrarily correlated and arbitrarily distributed random vector. For this, two filtering algorithms based on the criterion of linear minimum mean‐square error are proposed. The first algorithm is an optimal algorithm that can exactly compute the linear minimum mean‐square error estimate of system states. The second algorithm is a suboptimal algorithm that is proposed to reduce the computation and storage load of the proposed optimal algorithm. Computer simulations are carried out to evaluate the performance of the proposed algorithms. Copyright © 2013 John Wiley & Sons, Ltd.