Forward-backward Linear Prediction Research Articles

Packet Loss Concealment Estimating Residual Errors of Forward-Backward Linear Prediction for Bone-Conducted Speech

Packet Loss Concealment Estimating Residual Errors of Forward-Backward Linear Prediction for Bone-Conducted Speech

Parameter Estimation in Spectral Resolution Enhancement Based on Forward-Backward Linear Prediction Total Least Square Method.

In a Fourier transform infrared (IR) spectrometer, the Michelson interference signal extrapolation method based on linear prediction is often used to improve spectral resolution. In this method, an autoregressive (AR) model is established for the Michelson interference signal in the spectrometer. Once the AR model parameters are determined, the AR process is predictable. The interference signal can be used to figure out the AR model's parameters. Based on this, the AR model can be used to extrapolate the interference signal to improve the spectral resolution. In this paper, the forward-backward linear prediction total least squares (FB-TLS) method is proposed to estimate the parameters of the AR model. The parameters that are estimated are used to improve the IR spectral resolution. By simulating different order and signal-to-noise ratio situations, the effects of the Burg, the least square, and the FB-TLS parameter estimation methods on spectral resolution enhancement are studied. The simulation results demonstrate that the FB-TLS parameter estimation method can effectively suppress noise and avoid spurious peaks. The experimental results demonstrate that the FB-TLS parameter estimation method is effective for spectral resolution enhancement technology based on linear prediction. When the FB-TLS method is used to enhance NH3 IR spectral resolution from 2 cm-1 to 1 cm-1, the spectral prediction error in the NH3 characteristic band is only 0.21% compared with the measured NH3 spectrum, whose spectral resolution is 1 cm-1.

Formant Tracking Using Quasi-Closed Phase Forward-Backward Linear Prediction Analysis and Deep Neural Networks

Formant tracking is investigated in this study by using trackers based on dynamic programming (DP) and deep neural nets (DNNs). Using the DP approach, six formant estimation methods were first compared. The six methods include linear prediction (LP) algorithms, weighted LP algorithms and the recently developed quasi-closed phase forward-backward (QCP-FB) method. QCP-FB gave the best performance in the comparison. Therefore, a novel formant tracking approach, which combines benefits of deep learning and signal processing based on QCP-FB, was proposed. In this approach, the formants predicted by a DNN-based tracker from a speech frame are refined using the peaks of the all-pole spectrum computed by QCP-FB from the same frame. Results show that the proposed DNN-based tracker performed better both in detection rate and estimation error for the lowest three formants compared to reference formant trackers. Compared to the popular Wavesurfer, for example, the proposed tracker gave a reduction of 29%, 48% and 35% in the estimation error for the lowest three formants, respectively.

Spread Sea Clutter Suppression in HF Hybrid Sky-Surface Wave Radars Based on General Parameterized Time-Frequency Analysis

In high-frequency (HF) hybrid sky-surface wave radar, the first-order sea clutter broadening is severe under the action of ionospheric phase disturbance and bistatic angles. In this paper, a cascaded method is described to suppress the spread sea clutter. Firstly, the radar configuration and sea clutter broadening model are introduced based on the newly developed integrated HF sky-surface wave experimental system. In the cascaded processing method, a new ionospheric decontamination method based on general parameterized time-frequency (GPTF) analysis is proposed to estimate or correct the ionospheric phase distortion with large amplitude. Then, the forward-backward linear prediction (FBLP) algorithm is used to suppress the spread sea clutter caused by bistatic angle. Simulation results show that such ionospheric decontamination method based on GPTF is helpful for the large-amplitude ionospheric contamination when the target masking effect happens even after ionospheric phase decontamination. Finally, the proposed method is examined by the measured data. Experimental results indicate that the proposed method can well suppress the broadening sea clutter for HF hybrid sky-surface wave radars.

Time-Delay Estimation Using Ground-Penetrating Radar With a Support Vector Regression-Based Linear Prediction Method

Ground-penetrating radars (GPR) are widely used in media parameters’ estimation and targets’ localization. This paper focuses on time-delay estimation (TDE) using the GPR signal, which contains important information about the probed media structure. However, TDE tends to be a challenging task in GPR applications, in the scenarios of overlapping, coherent signals and limited snapshots. Forward–backward linear prediction (FBLP) is a high time-resolution method, which is able to directly deal with coherent signals. Support vector regression (SVR) is robust with small samples. Therefore, we propose to combine the theory of FBLP and SVR together to enhance the robustness of TDE in the case of coherent, overlapping signals as well as limited snapshots. The proposed method is tested with both numerical and experimental data. Both the results demonstrate the effectiveness of the proposed method.

A hybrid method for MEMS gyroscope signal error compensation

PurposeThis paper aims to propose a hybrid method based on polynomial fitting bias self-compensation, grey forward-backward linear prediction (GFBLP) and moving average filter (MAF) for error compensation in micro-electromechanical system gyroscope signal especially under motion state.Design/methodology/approachThe error compensation can be divided into two processes: bias correction and noise reduction. A polynomial drift angle fitting algorithm is used to correct bias before denoising processing. For noise reduction, operation can be taken in two stages: detection and processing. First, sample variances are used to judge motion state. According to the detection results, algorithmic system switches between grey GFBLP and MAF to ensure fast convergence rate and small steady-state error.FindingsExperimental results show that the proposed method can correct bias effectively for practical gyroscope signal, and can eliminate noise effectively for both practical gyroscope signal and synthetic signal, which indicates the effectiveness of the proposed method.Originality/valueBias correction and noise reduction are considerations. Noise contained in practical or synthetic signal can be reduced rapidly and effectively, which benefits from the new idea of combination grey GFBLP, MAF and sample variances. And most importantly, it is applicable for signal denoising under arbitrary motion state condition, which is different from other methods where the convergence performance is seldom analyzed.

Adaptive Subspace Method for Phase Noise Estimation in CO-OFDM Systems

In this paper, a blind phase noise estimation method for a coherent optical orthogonal frequency-division multiplexing (CO-OFDM) system is considered. Based on the subspace approach, a simple and robust method is proposed to adaptively estimate and track phase noise in CO-OFDM systems. The idea behind this novel technique is to estimate the singular vectors that correspond to the smallest singular values of the noise subspace. A weighting parameter, which is derived based on the forward–backward linear prediction technique, is subsequently constructed using the obtained singular vectors of the noise subspace to adaptively estimate the phase noise in the system. Also, a variable step size is introduced to ensure an improved performance as well as stable convergence. Simulation results are shown to demonstrate the effectiveness as well as the efficiency of the proposed algorithm, which is implemented in a polarization multiplexed CO-OFDM system stressed by polarization mode dispersion, chromatic dispersion, and other polarization-dependent losses along the fiber link.

Hybrid methods for MEMS gyro signal noise reduction with fast convergence rate and small steady-state error

Abstract In this paper, a hybrid method is proposed for noise reduction in MEMS gyro signal. To ensure rapid response rate and small steady-state error, and by simultaneously considering the motion state complexity of noisy signal especially under dynamic state, denoising scheme is well-designed, which can be divided into three steps: distinguishing different IMFs modes, determining current motion state, and selecting proper denoising method. Two carefully selected indexes divide the IMFs into three parts, noisy IMFs, mixed IMFs and information IMFs, with the mixed IMFs needed further processing. Sample variances based on AMA are used to determine current motion state. Accordingly, soft interval thresholding, soft thresholding, or forward-backward linear prediction is selected to reduce noise components contained in the mixed IMFs. Denoised mixed IMFs and information IMFs constitute final denoised signal. Practical MEMS gyro signal under different motion conditions are employed to validate the effectiveness of the proposed method. Hilbert spectral analysis and Allan variance further verify the proposed method from qualitative and quantitative point of view. Besides, computational time complexity is also analyzed.

Quasi-closed phase forward-backward linear prediction analysis of speech for accurate formant detection and estimation

Recently, a quasi-closed phase (QCP) analysis of speech signals for accurate glottal inverse filtering was proposed. However, the QCP analysis which belongs to the family of temporally weighted linear prediction (WLP) methods uses the conventional forward type of sample prediction. This may not be the best choice especially in computing WLP models with a hard-limiting weighting function. A sample selective minimization of the prediction error in WLP reduces the effective number of samples available within a given window frame. To counter this problem, a modified quasi-closed phase forward-backward (QCP-FB) analysis is proposed, wherein each sample is predicted based on its past as well as future samples thereby utilizing the available number of samples more effectively. Formant detection and estimation experiments on synthetic vowels generated using a physical modeling approach as well as natural speech utterances show that the proposed QCP-FB method yields statistically significant improvements over the conventional linear prediction and QCP methods.

DOA Finding with Support Vector Regression Based Forward-Backward Linear Prediction.

Direction-of-arrival (DOA) estimation has drawn considerable attention in array signal processing, particularly with coherent signals and a limited number of snapshots. Forward–backward linear prediction (FBLP) is able to directly deal with coherent signals. Support vector regression (SVR) is robust with small samples. This paper proposes the combination of the advantages of FBLP and SVR in the estimation of DOAs of coherent incoming signals with low snapshots. The performance of the proposed method is validated with numerical simulations in coherent scenarios, in terms of different angle separations, numbers of snapshots, and signal-to-noise ratios (SNRs). Simulation results show the effectiveness of the proposed method.

Algorithmic Enhancement of Spectral Resolution of a LiNbO3 Waveguide-Based Miniature Fourier Transform Spectrometer.

In a recent report we demonstrated a miniature static Fourier transform spectrometer (FTS) that was implemented with a LiNbO3 (LN) waveguide electro-optic modulator (EOM) combined with the dispersion relation between its half-wave voltage and wavelength. The FTS was verified to be able to measure laser wavelength and for low-resolution spectroscopy. In this report, we successfully applied the resolution enhancement algorithm to the FTS, resulting in at least a three-fold increase in its spectral resolution without causing obvious distortion of the measured spectra. The algorithm method used is based on an autoregressive (AR) model, singular value decomposition (SVD), and forward-backward linear prediction (FBLP). The combination of these methods allows the FTS to remain a small size but to possess good spectral resolution, effectively mitigating the conflict between the small size and high resolution of the device. This study opens the way to development of high-resolution miniature FTS.

Cascaded method for ionospheric decontamination and sea clutter suppression for high‐frequency hybrid sky‐surface wave radar

In high-frequency (HF) hybrid sky-surface wave radar, the first-order sea clutter broadening is very severely under the influence of ionosphere and bistatic angle, which affects the detection of ship target. In this study, a cascaded method for ionospheric decontamination and sea clutter suppression to enable detection of ship target is presented. Firstly, the radar configuration, sea clutter broadening characteristics, and model are introduced based on the newly-developed integrated HF sky-surface wave experimental system. In the cascaded processing method, the time-frequency analysis method based on S 2 -method is used to correct the ionospheric phase contamination with large amplitude. Then, a novel forward-backward linear prediction algorithm is proposed to suppress the broadening bistatic sea clutter caused by bistatic angle, which is based on the distribution characteristics and the multi-dimensional feature of first-order sea clutter. Finally, the proposed method is examined by measured data. Experimental results indicate that the proposed method can well suppress the broadening sea clutter and improve the ship target detection performance.

A Novel DOA Estimation Approach for Unknown Coherent Source Groups with Coherent Signals

In this paper, a new combination of Minimum Description Length (MDL) or Eigenvalue Gradient Method (EGM), Joint Approximate Diagonalization of Eigenmatrices (JADE) and Modified Forward-Backward Linear Prediction (MFBLP) algorithms is proposed which determines the number of non-coherent source groups and estimates the Direction Of Arrivals (DOAs) of coherent signals in each group. First, the MDL/EGM algorithm determines the number of non-coherent signal groups, and then the JADE algorithm estimates the generalized steering vectors considering white/colored Gaussian noise. Finally, the MFBLP algorithm is applied to estimate DOAs of coherent signals in each group. A new Normalized Root Mean Square Error (NRMSE) is also proposed that introduces a more realistic metric to compare the performance of DOA estimation methods. Simulation results show that the proposed algorithm can resolve sources regardless of QAM modulation size. In addition, simulations in white/colored Gaussian noises show that the proposed algorithm outperforms the JADE-MUSIC algorithm in a wide range of Signal to Noise Ratios (SNRs).

Noise-Reduced Complex LPC Analysis for Formant Estimation of Noisy Speech

In this paper we present a complex linear prediction analysis method for estimating the formant frequencies of noisy speech. The proposed method effectively utilizes the signal (being the analytic signal) which is ignored in the conventional complex linear prediction analysis to achieve noise reduction. Also, the covariance and forward-backward linear prediction (FBLP) methods are compared, and the FBLP method is deployed for predictive coefficients estimation in the proposed method. Experimental results show that the proposed method yields better performance of formant estimation when compared to some conventional methods in white noise. 

NMR Conformational and Dynamic Consequences of a Gly to Ser Substitution in an Osteogenesis Imperfecta Collagen Model Peptide

Close packing of three chains in a standard collagen triple helix requires Gly as every third residue. Missense mutations replacing one Gly by a larger residue in the tripeptide repeating sequence in type I collagen are common molecular causes of osteogenesis imperfecta. The structural and dynamic consequences of such mutations are addressed here by NMR studies on a peptide with a Gly-to-Ser substitution within an alpha1(I) sequence. Distances derived from nuclear Overhauser effects indicate that the three Ser residues are still packed in the center of the triple helix and that the standard 1-residue stagger is maintained. NMR dynamics using H-exchange and temperature-dependent amide chemical shifts indicate a greater disruption of hydrogen bonding and/or increased conformational flexibility C-terminal to the Ser site when compared with N terminal. This is consistent with recent suggestions relating clinical severity with an asymmetric effect of residues N- versus C-terminal to a mutation site. Dynamic studies also indicate that the relative position between a Gly in one chain and the mutation site in a neighboring staggered chain influences the disruption of the standard hydrogen-bonding pattern. The structural and dynamic alterations reported here may play a role in the etiology of osteogenesis imperfecta by affecting collagen secretion or interactions with other matrix molecules.

Study of microarray time series data based on Forward-Backward Linear Prediction and Singular Value Decomposition

We propose a method to analyse the periodicities of gene expression profiles based on the spectral domain approach. Our spectral reconstruction method outperforms three other recently proposed methods, which do not require any prior knowledge. It is proven that an alternative method for studying cell-cycle regulation is possible even where very little prior knowledge is available. We also investigate the potential of combining signals with similar frequency components to form an overdetermined system of equations, and use least squares solution to estimate the spectral frequency. Results show that this new method is able to estimate the peak frequency more accurately.

OTHR Impulsive Interference Suppression in Strong Clutter Background

External interferences can severely degrade the performance of an Over-the-horizon radar (OTHR), so suppression of external interferences in strong clutter environment is the prerequisite for the target detection. The traditional suppression solutions usually began with clutter suppression in either time or frequency domain, followed by the interference detection and suppression. Based on this traditional solution, this paper proposes a method characterized by joint clutter suppression and interference detection: by analyzing eigenvalues in a short-time moving window centered at different time position, clutter is suppressed by discarding the maximum three eigenvalues at every time position and meanwhile detection is achieved by analyzing the remained eigenvalues at different position. Then, restoration is achieved by forward-backward linear prediction using interference-free data surrounding the interference position. In the numeric computation, the eigenvalue decomposition (EVD) is replaced by singular values decomposition (SVD) based on the equivalence of these two processing. Data processing and experimental results show its efficiency of noise floor falling down about 10-20dB.

Multi-scale parametric spectral analysis for exon detection in DNA sequences based on forward-backward linear prediction and singular value decomposition of the double-base curves

This paper presents a new method for exon detection in DNA sequences based on multi-scale parametric spectral analysis. A forward-backward linear prediction (FBLP) with the singular value decomposition (SVD) algorithm FBLP-SVD is applied to the double-base curves (DB-curves) of a DNA sequence using a variable moving window sizes to estimate the signal spectrum at multiple scales. Simulations are done on short human genes in the range of 11bp to 2032bp and the results show that our proposed method out-performs the classical Fourier transform method. The multi-scale approach is shown to be more effective than using a single scale with a fixed window size. In addition, our method is flexible as it requires no training data.

An Adaptive Predictor With Cascaded Forward-Backward Structure

Further developing the idea of a cascade structure for adaptive linear prediction with independently adapting low-order stages, we develop a new implementation where the single stages use the forward-backward linear prediction algorithm. Combining the advantages of the cascade structure, which is less affected by the eigenvalue spread and mode coupling, with the reduced misadjustment of the forward-backward predictor delivers a new structure with improved performance. This new structure is capable of faster and more accurate tracking of nonstationary inputs in the early stage of the adaptation.

Contour Estimation by Array Processing Methods

This work is devoted to the estimation of rectilinear and distorted contours in images by high-resolution methods. In the case of rectilinear contours, it has been shown that it is possible to transpose this image processing problem to an array processing problem. The existing straight line characterization method called subspace-based line detection (SLIDE) leads to models with orientations and offsets of straight lines as the desired parameters. Firstly, a high-resolution method of array processing leads to the orientation of the lines. Secondly, their offset can be estimated by either the well-known method of extension of the Hough transform or another method, namely, the variable speed propagation scheme, that belongs to the array processing applications field. We associate it with the method called "modified forward-backward linear prediction" (MFBLP). The signal generation process devoted to straight lines retrieval is retained for the case of distorted contours estimation. This issue is handled for the first time thanks to an inverse problem formulation and a phase model determination. The proposed method is initialized by means of the SLIDE algorithm.