Showing papers on "Kernel adaptive filter published in 2022"

Optimal Adaptive Filtering Algorithm by Using the Fractional-Order Derivative

Abstract: The previous work for the filter design considers uncorrelated white measurement noise disturbance. For more complex correlated noise disturbance, the conventional adaptive filter results in biased estimates. To overcome this problem, we introduce a linear prefilter to whiten the correlated noise (i.e., colored noise) for obtaining the unbiased estimate of the filter weight. Moreover, the design of some adaptive filters mainly focuses on the integer-order optimization methods. However, compared with the integer-order-based adaptive algorithms, the fractional-order-based algorithms show better performance. Thus, this letter develops a new gradient approach for the adaptive filter design based on the fractional-order derivative and a linear filter. Finally, the simulation results are provided from the system identification perspective for demonstrating the performance analysis of the proposed algorithms.

Modified Proportionate Affine Projection Algorithm Based Adaptive DSTATCOM Control With Increased Convergence Speed

Abstract: The three-phase DSTATCOM is prone to higher dynamics due to grid disturbances. The proportionate affine projection algorithm (PAPA) is an adaptive filter that can be used for DSTATCOM control. In order to overcome the sluggish convergence speed of adaptive filters, PAPA is proposed in this paper. The convergence rate versus the steady-state error is a trade-off in conventional adaptive filters. However, the utilization of two adaptive filters in CSS-PAPA increases the convergence and decreases the steady-state error. The suggested filter has the advantage of having a lower computational cost than a standard adaptive filter. The proposed filter helps the inverter to work as a shunt compensator. The goal of the suggested controller is to adjust for reactive power and unity power factor during faulty conditions. The proposed DSTATCOM controller has undergone experimental validation in the laboratory.

Recursive Optimal Finite Impulse Response Filter and Its Application to Adaptive Estimation

Abstract: In this paper, the recursive form of an optimal finite impulse response filter is proposed for discrete time-varying state-space models. The recursive form of the finite impulse response filter is derived by employing finite horizon Kalman filtering with optimally estimated initial conditions. The horizon initial state and its error covariance on the horizon are optimally estimated by using recent finite measurements, in the sense of maximum likelihood estimation, then initiating the finite horizon Kalman filter. The optimality and unbiasedness of the proposed filter are proved by comparison with the conventional optimal finite impulse response filter in batch form. Moreover, an adaptive FIR filter is also proposed by applying the adaptive estimation scheme to the proposed recursive optimal FIR filter as its application. To evaluate the performance of the proposed algorithms, a computer simulation is performed to compare the conventional Kalman filter and adaptive Kalman filters for the gas turbine aircraft engine model.

Generalized Maximum Complex Correntropy Augmented Adaptive IIR Filtering

Abstract: Augmented IIR filter adaptive algorithms have been considered in many studies, which are suitable for proper and improper complex-valued signals. However, lots of augmented IIR filter adaptive algorithms are developed under the mean square error (MSE) criterion. It is an ideal optimality criterion under Gaussian noises but fails to model the behavior of non-Gaussian noise found in practice. Complex correntropy has shown robustness under non-Gaussian noises in the design of adaptive filters as a similarity measure for the complex random variables. In this paper, we propose a new augmented IIR filter adaptive algorithm based on the generalized maximum complex correntropy criterion (GMCCC-AIIR), which employs the complex generalized Gaussian density function as the kernel function. Stability analysis provides the bound of learning rate. Simulation results verify its superiority.

Student’s t-Kernel-Based Maximum Correntropy Kalman Filter

Abstract: The state estimation problem is ubiquitous in many fields, and the common state estimation method is the Kalman filter. However, the Kalman filter is based on the mean square error criterion, which can only capture the second-order statistics of the noise and is sensitive to large outliers. In many areas of engineering, the noise may be non-Gaussian and outliers may arise naturally. Therefore, the performance of the Kalman filter may deteriorate significantly in non-Gaussian noise environments. To improve the accuracy of the state estimation in this case, a novel filter named Student’s t kernel-based maximum correntropy Kalman filter is proposed in this paper. In addition, considering that the fixed-point iteration method is used to solve the optimal estimated state in the filtering algorithm, the convergence of the algorithm is also analyzed. Finally, comparative simulations are conducted and the results demonstrate that with the proper parameters of the kernel function, the proposed filter outperforms the other conventional filters, such as the Kalman filter, Huber-based filter, and maximum correntropy Kalman filter.

Neural-Adaptive Stochastic Attitude Filter on SO(3)

Abstract: This letter proposes a novel stochastic nonlinear neural-adaptive-based filter on $SO(3)$ for the attitude estimation problem. The proposed filter produces good results given measurements extracted from low-cost sensing units (e.g., IMU or MARG sensor modules). The filter is guaranteed to be almost semi-globally uniformly ultimately bounded in the mean square. In addition to Lie Group formulation, quaternion representation of the proposed filter is provided. The effectiveness of the proposed neural-adaptive filter is tested and evaluated in its discrete form under the conditions of large initialization error and high measurement uncertainties.

Low Complexity RLS Adaptive Filters

Abstract: This paper presents two adaptive filters with a reduced arithmetic complexity which are based on the Recursive Least Squares (RLS) algorithms. The first one is the cascaded adaptive filter. The second one is the adaptive filter with the diagonalized correlation matrix of the input signal. The both filters have a reduced arithmetic complexity comparing to the direct implementation of the adaptive filter. The cost of the reduction is some degradation of the adaptive filter performance. The reduction is achieved only if the RLS algorithms with quadratic complexity are used. The computational procedures and the arithmetic complexities of the considered adaptive filters are the same, but the performance is different. This paper presents the RLS algorithms based on the Matrix Inversion Lemma (MIL). However, all the results and conclusions are valid for any RLS algorithms with quadratic complexity. The paper demonstrates the considered adaptive filter performance via simulation.

Neural-Adaptive Stochastic Attitude Filter on SO(3)

Abstract: This letter proposes a novel stochastic nonlinear neural-adaptive-based filter on $SO(3)$ for the attitude estimation problem. The proposed filter produces good results given measurements extracted from low-cost sensing units (e.g., IMU or MARG sensor modules). The filter is guaranteed to be almost semi-globally uniformly ultimately bounded in the mean square. In addition to Lie Group formulation, quaternion representation of the proposed filter is provided. The effectiveness of the proposed neural-adaptive filter is tested and evaluated in its discrete form under the conditions of large initialization error and high measurement uncertainties.

Parallel APSM for Fast and Adaptive Digital SIC in Full-Duplex Transceivers with Nonlinearity

Abstract: This paper presents a kernel-based adaptive filter that is applied for the digital domain self-interference cancellation (SIC) in a transceiver operating in full-duplex (FD) mode. In FD, the benefit of simultaneous transmission and receiving of signals comes at the price of strong self-interference (SI). In this work, we are primarily interested in suppressing the SI using an adaptive filter namely adaptive projected subgradient method (APSM) in a reproducing kernel Hilbert space (RKHS) of functions. Using the projection concept as a powerful tool, APSM is used to model and consequently remove the SI. A low-complexity and fast-tracking algorithm is provided taking advantage of parallel projections as well as the kernel trick in RKHS. The performance of the proposed method is evaluated on real measurement data. The method illustrates the good performance of the proposed adaptive filter, compared to the known popular benchmarks. They demonstrate that the kernel-based algorithm achieves a favorable level of digital SIC while enabling parallel computation-based implementation within a rich and nonlinear function space, thanks to the employed adaptive filtering method.

Kernel Filter Proportionate Normalized Least Mean Square Algorithm for Nonlinear Sparse Systems

Abstract: An effective nonparametric model for creating adaptive nonlinear filtering (ANF) algorithms is provided by kernel methods. These algorithms were developed based on different kernels like Gaussian and Laplacian. Moreover, in practical applications, nonlinear systems are also sparse in nature. So, the filter proportionate normalized least mean square (FPNLMS) algorithm was developed for such sparse nonlinear systems. In this paper to have an effective ANF algorithm, the Gaussian kernel method is used with the FPNLMS algorithm providing a novel kernel FPNLMS (KFPNLMS) algorithm for sparse environments is developed. The system identification problem is solved using the KFPNLMS method, which performs as expected when the convergence analysis is done.

Implementation of FIR Adaptive Filter with FPGA

Abstract: The trend of using adaptive filtering for an emerging communication network is increasing rapidly, because adaptive algorithms used in adaptive filters can intelligently adapt in the cases of time-varying channels, to obtain the best possible outcome. Various algorithms are used for adaptive filtering. In this paper, the Least Mean Square algorithm used to design FIR adaptive filter is synthesized on Xilinx ISE. No of slice registers, no of Flip flops, and LUTS are calculated through the Synthesis report.RTL view are also observed here.Test bench is also generated for simulation. Efficient hardware resource utilization of the proposed design of adaptive FIR filter are observed here.

A new approach for SPN removal: nearest value based mean filter

Abstract: In this study, a new adaptive filter is proposed to eliminate salt and pepper noise (SPN). The basis of the proposed method consists of two-stages. (1) Changing the noisy pixel value with the closest pixel value or assigning their average to the noisy pixel in case there is more than one pixel with the same distance; (2) the updating of the calculated noisy pixel values with the average filter by correlating them with the noise ratio. The method developed was named as Nearest Value Based Mean Filter (NVBMF), because of using the pixel value which the closest distance in the first stage. Results obtained with the proposed method: it has been compared with the results obtained with the Adaptive Frequency Median Filter, Adaptive Riesz Mean Filter, Improved Adaptive Weighted Mean Filter, Adaptive Switching Weight Mean Filter, Adaptive Weighted Mean Filter, Different Applied Median Filter, Iterative Mean Filter, Two-Stage Filter, Multistage Selective Convolution Filter, Different Adaptive Modified Riesz Mean Filter, Stationary Framelet Transform Based Filter and A New Type Adaptive Median Filter methods. In the comparison phase, nine different noise levels were applied to the original images. Denoised images were compared using Peak Signal-to-Noise Ratio, Image Enhancement Factor, and Structural Similarity Index Map image quality metrics. Comparisons were made using three separate image datasets and Cameraman, Airplane images. NVBMF achieved the best result in 52 out of 84 comparisons for PSNR, best in 47 out of 84 comparisons for SSIM, and best in 36 out of 84 comparisons for IEF. In addition, values nearly to the best result were obtained in comparisons where the best result could not be reached. The results obtained show that the NVBMF can be used as an effective method in denoising SPN.

A Fast Adaptive Multi-Scale Kernel Correlation Filter Tracker for Rigid Object

Abstract: The efficient and accurate tracking of a target in complex scenes has always been one of the challenges to tackle. At present, the most effective tracking algorithms are basically neural network models based on deep learning. Although such algorithms have high tracking accuracy, the huge number of parameters and computations in the network models makes it difficult for such algorithms to meet the real-time requirements under limited hardware conditions, such as embedded platforms with small size, low power consumption and limited computing power. Tracking algorithms based on a kernel correlation filter are well-known and widely applied because of their high performance and speed, but when the target is in a complex background, it still can not adapt to the target scale change and occlusion, which will lead to template drift. In this paper, a fast multi-scale kernel correlation filter tracker based on adaptive template updating is proposed for common rigid targets. We introduce a simple scale pyramid on the basis of Kernel Correlation Filtering (KCF), which can adapt to the change in target size while ensuring the speed of operation. We propose an adaptive template updater based on the Mean of Cumulative Maximum Response Values (MCMRV) to alleviate the problem of template drift effectively when occlusion occurs. Extensive experiments have demonstrated the effectiveness of our method on various datasets and significantly outperformed other state-of-the-art methods based on a kernel correlation filter.

A Fast Kernel Least Mean Square Algorithm

Abstract: To deal with the problems in the nonlinear system, the kernel adaptive filter (KAF) was proposed by processing data in the reproducing kernel Hilbert space (RKHS). However, the kernel method dramatically improves the amount of calculation of the filter, which limits its application in practical problems. Furthermore, a critical factor in a large amount of KAF computation is due to its slow convergence speed, which requires a large amount of training data to participate in the calculation. If we can accelerate the convergence speed of KAF, the amount of training data can be reduced, thereby reducing the amount of KAF computation. This paper proposes a fast kernel least mean square algorithm (FAST-KLMS) by adaptively updating step size to address this issue. To verify the superiority of FAST-KLMS, we have applied it to the simulations of nonlinear channel equalization. The simulation results show that FAST-KLMS needs less training data to complete the convergence, which has improved the filtering performance of KAF.

Fast LMS-based Compound Algorithm for Modelling Long Memory Filters with Colored Signal

Abstract: A new adaptive modelling technique (AP-NCD) is suggested to increase the speed and accuracy of modelling the long memory filter with strong correlation signals by integrating the alternating projection principle with the nonlinear multi-degree of freedom algorithm (NLMDF). The basic idea is to partition a huge filter into a set of sub filters of length m = 2 by using the notion of alternating projection, and then iterate through each sub filter using the NCD method derived from the NLMDF algorithm. The convergence characteristics of the AP-NCD algorithm are investigated in this study, and the principle of parameter selection is derived. The simulation results demonstrate that the AP-NCD algorithm has a faster convergence time and accuracy than the NLMDF, TPLMS, and DCTLMS algorithms when the high correlation input signal is used to model the long memory filter (FIR).

Non-linear Adaptive Wiener Filter for Time Series Smoothing

Abstract: Abstract Often, in the analysis of time series and digital signals, a smoothing procedure is required to filter undesired random perturbations as noise and outliers in data. Among the most widely known techniques for time series smoothing, we have convolutional filters, simple exponential smoothing, triple exponential smoothing (Holt-Winters method) and linear adaptive filters, such as the Wiener filter. In this paper, we propose the NoLAW filter (Non-Linear Adaptive Wiener filter), a higher-order nonlinear extension for the adaptive Wiener filter, which is a linear statistical smoothing technique that assumes the hypothesis that the underlying series is corrupted by a zero mean, additive and independent Gaussian noise. Numerical experiments show that the proposed method is a computationally efficient and viable approach for filtering time series. Quantitative metrics show that the NoLAW filter is capable of producing better results than the usual linear Wiener filter, simple exponential smoothing and Holt-Winters method. Moreover, the computational cost of the proposed NoLAW is linear in the number of samples, which means that, asymptotically, it is equivalent to the regular Wiener filter.

Sharpening the Spectral Response of Polynomial Graph Filter by Using Amplitude Change Function

Abstract: In graph signal processing applications, it is desirable to design a polynomial graph filter with good frequency selective performance. This task can be done by designing a large-order frequency selective graph filter with very narrow transient band. However, this design often suffers from the numerical stability problem when filter coefficients are determined by least squares methods. To solve this problem, a two-stage design method is presented in this paper. In the first stage, a low-order graph filter with wide transient band is designed for avoiding numerical problem. In the second stage, the amplitude change function is used to sharpen the spectral response of low-order filter to get a filter with narrow transient band. The designed filter can be implemented by cascading several identical filters, so it is also suitable for distributed implementation. Finally, low-pass filter design examples are presented to illustrate the benefits of the proposed method.

Visual Object Tracking Based on Adaptive Background-Awareness and Spatial Constraint

Abstract: The correlation filter method is effective in visual tracking tasks, whereas it suffers from the boundary effect and filter degradation in complex situations, which can result in suboptimal performance. Aiming at the solving above problem, this study proposes an object tracking method with a discriminant correlation filter, which combines an adaptive background perception and a spatial dynamic constraint. In this method, an adaptive background-awareness strategy is used to filter the background information trained by the interference filter to improve the discriminability between the object and the background. In addition, the spatial regularization term is introduced, and the dynamic change of the real filter and the predefined spatial constraint template is used to optimize filter learning to enhance the spatial information capture ability of the filter model. Experiments on the OTB100, VOT2018, and TrackingNet standard datasets demonstrate that our method achieves favorable accuracy and success rates. Compared with the current popular correlation filter methods, the proposed method can still maintain stable tracking performance with a scene scale variation, complex background, motion blur, and fast motion.

Research on Embedded Software based on adaptive filtering algorithm

Abstract: Through the thorough exploration of the adaptive filter structure and the LMS adaptive filter algorithm, the filter performance of the adaptive filter algorithm can be clearly mastered. The solution formula of LMS algorithm is based on it, and DSP software programming and Matlab simulation programming methods are used to lay the foundation for the effective implementation of LMS algorithm. Therefore, based on the adaptive filtering algorithm, the embedded software simulation development system is analyzed to help the application of adaptive filtering theory.

Novel Improvement for Nonlinear Compatibility of Least Mean Square Adaptive Algorithm

Abstract: In order to improve the least mean squares (LMS) adaptation algorithm to accommodate the nonlinear transfer function, and to adjust the coefficients of adaptive filter during the actual implement of bias voltage and signal amplitude, methods are proposed and simulated to develop a nonlinear adaptive filter. The inputs to LMS are replaced by the derivatives of traditional inputs, and the step for each training iteration is adaptively controlled by the difference between target signal and actual signal. The simulation utilizes the implementation of Nyquist pulses optical sampling and works as a digital signal processing pre-compensation to reduce influence of the frequency responses on wires and devices. The simulation result shows promising improvement with the modified adaptation algorithm method in tackling Mach Zehnder modulator's non-monotonic transfer function.

Intelligent Filter of L1 Adaptive Controller Using ANFIS System Optimized with Genetic Algorithm

Abstract: $\mathcal{L}_{1}$ adaptive controller is known for ensuring fast adaptation with the optimal transient performance for input and output signals using a low-pass filter with adjustable gain in the feedback loop. During choosing a filter gain, the main criteria are the tradeoff between the performance, robustness and fast adaptation. There are several methods with different complexities for determining bounds or initial values of filter gain, such values may require the use of many iterations of trial and error implemented in the controller. In these methods, the specified gain is kept constant, which leads to non-optimal performance in adaptation speed and robustness. In this paper, a new approach based on an adaptive-neural-fuzzy inference system (ANFIS) optimized with a genetic algorithm is presented to continuously determine this gain and optimal performance according to the error rate. The objective function optimized by the genetic algorithm is also determined in terms of position tracking error, speed tracking error, and control signal. To evaluate the performance, the proposed method is implemented on a nonlinear feedback system in the presence of unmatched uncertainties. The simulation results show that the controller performance improves more than the constant filter efficiency in this method.

Vehicle Cabin Noise Cancellation Model Using Pre-Filter for Improved Convergence Rate and Better Stability

Abstract: Resume Adaptive algorithms are used in updating the filter coefficients for active noise cancellation applications in reduction of vehicle cabin noise. The performance of the adaptive algorithms in low-frequency noise cancellation depends on how efficiently it alters the filter coefficient to minimize the difference between the approximated signal and the original one. Here is proposed an active noise cancellation model, using the low pass fixed coefficient filter before the adaptive filter, in order to improve the performance of the adaptive algorithm. Convergence rate, SNR and error vector magnitude are analysed for of adaptive algorithm in support of our research results.

Adaptive particle filter for state estimation with application to non‐linear system

Abstract: Particle filtering (PF) has certain application value, but the disadvantage is that there is a phenomenon of particle degradation. In order to reduce the impact of this problem, this paper presents a new adaptive PF approach to improve the estimate accuracy. From the perspective of selecting an appropriate important density functions, in this filter, the particles are first updated using the Spherical Simplex Unscented Kalman Filter algorithm, and then the particles are updated using the Adaptive Extended Kalman filter algorithm. Simultaneously, from the perspective of improving the resampling method, a new resampling technique based on the random resampling method has been designed and fused to this filter. The comparison and analysis of two simulation schemes have been conducted to assess the performance of the designed filtering algorithm. The simulation results show the effectiveness of the proposed approach.

ProNet: Adaptive Process Noise Estimation for INS/DVL Fusion

Abstract: Inertial and Doppler velocity log sensors are commonly used to provide the navigation solution for autonomous underwater vehicles (AUV). To this end, a nonlinear filter is adopted for the fusion task. The filter's process noise covariance matrix is critical for filter accuracy and robustness. While this matrix varies over time during the AUV mission, the filter assumes a constant matrix. Several models and learning approaches in the literature suggest tuning the process noise covariance during operation. In this work, we propose ProNet, a hybrid, adaptive process, noise estimation approach for a velocity-aided navigation filter. ProNet requires only the inertial sensor reading to regress the process noise covariance. Once learned, it is fed into the model-based navigation filter, resulting in a hybrid filter. Simulation results show the benefits of our approach compared to other models and learning adaptive approaches.