arXiv subjects

Wei-Ping Zhu

Publications and source records attributed to Wei-Ping Zhu.

36 records · Page 2Linked to original sources

arXiv2021

TSTNN: Two-stage Transformer based Neural Network for Speech Enhancement in the Time Domain

In this paper, we propose a transformer-based architecture, called two-stage transformer neural network (TSTNN) for end-to-end speech denoising in the time domain. The proposed model is composed of an encoder, a two-stage transformer module (TSTM), a masking module and a decoder. The encoder maps input noisy speech into feature representation. The TSTM exploits four stacked two-stage transformer blocks to efficiently extract local and global information from the encoder output stage by stage. The masking module creates a mask which will be multiplied with the encoder output. Finally, the decoder uses the masked encoder feature to reconstruct the enhanced speech. Experimental results on the benchmark dataset show that the TSTNN outperforms most state-of-the-art models in time or frequency domain while having significantly lower model complexity.

eess.AS↗

arXiv2021

Atomic Norm Based Localization of Far-Field and Near-Field Signals with Generalized Symmetric Arrays

Most localization methods for mixed far-field (FF) and near-field (NF) sources are based on uniform linear array (ULA) rather than sparse linear array (SLA). In this paper, we propose a localization method for mixed FF and NF sources based on the generalized symmetric linear arrays, which include ULAs, Cantor array, Fractal array and many other SLAs. Our method consists of two steps. In the first step, the high-order statistics of the array output is exploited to increase the degree of freedom. Then the direction-of-arrivals (DOAs) of the FF and NF sources are jointly estimated by using the recently proposed atomic norm minimization (ANM), which belongs to the gridless super-resolution method since the discretization of the parameter space is not required. In the second step, the ranges are given by MUSIC-like one-dimensional searching. Simulations results are provided to demonstrate the advantages of our method.

cs.IT↗

arXiv2021

Bit Error Rate Analysis for Reconfigurable Intelligent Surfaces with Phase Errors

In this paper, we analyze the error probability of reconfigurable intelligent surfaces (RIS)-enabled communication systems with quantized channel phase compensation over Rayleigh fading channels. The probability density and characteristic functions of the received signal amplitude are derive dand used to compute exact expressions for the bit error rate(BER). The resulting expressions are general, as they hold for an arbitrary number of reflecting elements N, and quantization levels, L. We introduce an exact asymptotic analysis in the high signal-to-noise ratio (SNR) regime, from which we demonstrate, in particular, that the diversity order is N/2 when L=2 and N when L >2. The theoretical frameworks and findings are validated with the aid of Monte Carlo simulation.

cs.IT↗

arXiv2020

Performance Evaluation and Diversity Analysis of RIS-Assisted Communications Over Generalized Fading Channels in the Presence of Phase Noise

In this paper, we develop a comprehensive theoretical framework for analyzing the performance of reconfigurable intelligent surfaces (RISs)-assisted communication systems over generalized fading channels and in the presence of phase noise. To this end, we propose the Fox's H model as a unified fading distribution for a large number of widely used generalized fading channels. In particular, we derive a unified analytical framework for computing the outage probability and for estimating the achievable diversity order of RIS-aided systems in the presence of phase shifts that either are optimally configured or are impaired by phase noise. The resulting expressions are general, as they hold for an arbitrary number of reflecting elements, and various channel fading and phase noise distributions. As far as the diversity order is concerned, notably, we introduce an asymptotic analytical framework for determining the diversity order in the absence of phase noise, as well as sufficient conditions based on upper bounds and lower bounds for ensuring that RIS-assisted systems achieve the full diversity order in the presence of phase noise. More specifically, if the absolute difference between pairs of phase errors is less than $π/2$, RIS-assisted communications achieve the full diversity order over independent fading channels, even in the presence of phase noise. The theoretical frameworks and findings are validated with the aid of Monte Carlo simulations.

cs.IT↗

arXiv2020

Tensor-Decomposition-based Hybrid Beamforming Design for mmWave OFDM Massive MIMO Communications

In this paper, we propose a novel joint hybrid precoder and combiner design for maximizing the average achievable sum-rate of single-user orthogonal frequency division multiplexing millimeter wave massive MIMO systems. We formulate the analog precoder and combiner design as a constrained Tucker2 decomposition and solve it by using the projected alternate least square method. Such a formulation allows maximizing the sum of the effective baseband gains over all subcarriers, while suppressing the interference among the data streams in the same subcarrier. In turn, the digital precoder and combiner are obtained from the effective baseband channel's singular value decomposition on a per-subcarrier basis. Numerical simulation results show that the proposed method outperforms other existing designs.

cs.IT↗

arXiv2020

On the Use of Audio Fingerprinting Features for Speech Enhancement with Generative Adversarial Network

The advent of learning-based methods in speech enhancement has revived the need for robust and reliable training features that can compactly represent speech signals while preserving their vital information. Time-frequency domain features, such as the Short-Term Fourier Transform (STFT) and Mel-Frequency Cepstral Coefficients (MFCC), are preferred in many approaches. While the MFCC provide for a compact representation, they ignore the dynamics and distribution of energy in each mel-scale subband. In this work, a speech enhancement system based on Generative Adversarial Network (GAN) is implemented and tested with a combination of Audio FingerPrinting (AFP) features obtained from the MFCC and the Normalized Spectral Subband Centroids (NSSC). The NSSC capture the locations of speech formants and complement the MFCC in a crucial way. In experiments with diverse speakers and noise types, GAN-based speech enhancement with the proposed AFP feature combination achieves the best objective performance while reducing memory requirements and training time.

eess.AS↗

arXiv2020

A Comprehensive Study of Reconfigurable Intelligent Surfaces in Generalized Fading

Leveraging on the reconfigurable intelligent surface (RIS) paradigm for enabling the next Internet of Things (IoT) and 6G era, this paper develops a comprehensive theoretical framework characterizing the performance of RIS-assisted communications in a plethora of propagation environments. We derive unified mathematical models for the outage probability and ergodic capacity of single and multiple-element RIS over Fox's H fading channel, which includes as special cases nearly all linear and non linear multi-path and shadowing fading models adopted in the open literature. For gleaning further insights, we capitalize on the algebraic asymptotic expansions of the H-transform to further analyze the outage probability and capacity at high signal-to-noise ratio (SNR) in a unified fashion. Asymptotic analysis shows two scaling rates of the outage probability at large average SNR. Moreover, by harnessing its tractability, the developed statistical machinery is employed to characterize the performance of multiple randomly distributed RIS-assisted communications over Fox's H fading channels. We show that there is a great potential to improve the outage performance and thereby the capacity when fewer RISs are deployed each with more reflecting elements.

eess.SP↗

arXiv2020

Genetic Algorithm Optimized Support Vector Machine in NOMA-Based Satellite Networks with Imperfect CSI

With the help of a power-domain non-orthogonal multiple access (NOMA) scheme, satellite networks can simultaneously serve multiple users within limited time/spectrum resource block. However, the existence of channel estimation errors inevitably degrade the judgment on users' channel state information (CSI) accuracy, thus affecting the user pairing processing and suppressing the superiority of the NOMA scheme. Inspired by the advantages of machine learning (ML) algorithms, we propose an improved support vector machine (SVM) scheme to reduce the inappropriate user pairing risks and enhance the performance of NOMA based satellite networks with imperfect CSI. Particularly, a genetic algorithm (GA) is employed to optimize the regularization and kernel parameters of the SVM, which effectively improves the classification accuracy of the proposed scheme. Simulations are provided to demonstrate that the performance of the proposed method is better than that with random user paring strategy, especially in the scenario with a large number of users.

eess.SP↗

arXiv2019

S-ConvNet: A Shallow Convolutional Neural Network Architecture for Neuromuscular Activity Recognition Using Instantaneous High-Density Surface EMG Images

The concept of neuromuscular activity recognition using instantaneous high-density surface electromyography (HD-sEMG) images opens up new avenues for the development of more fluid and natural muscle-computer interfaces. However, the existing approaches employed a very large deep convolutional neural network (ConvNet) architecture and complex training schemes for HD-sEMG image recognition, which requires the network architecture to be pre-trained on a very large-scale labeled training dataset, as a result, it makes computationally very expensive. To overcome this problem, we propose S-ConvNet and All-ConvNet models, a simple yet efficient framework for learning instantaneous HD-sEMG images from scratch for neuromuscular activity recognition. Without using any pre-trained models, our proposed S-ConvNet and All-ConvNet demonstrate very competitive recognition accuracy to the more complex state of the art for neuromuscular activity recognition based on instantaneous HD-sEMG images, while using a ~ 12 x smaller dataset and reducing learning parameters to a large extent. The experimental results proved that the S-ConvNet and All-ConvNet are highly effective for learning discriminative features for instantaneous HD-sEMG image recognition especially in the data and high-end resource constrained scenarios.

eess.SP↗

arXiv2018

Joint Doppler and Channel Estimation with Nested Arrays for Millimeter Wave Communications

Channel estimation is essential for precoding/combining in millimeter wave (mmWave) communications. However, accurate estimation is usually difficult because the receiver can only observe the low-dimensional projection of the received signals due to the hybrid architecture. We take the high speed scenario into consideration where the Doppler effect caused by fast-moving users can seriously deteriorate the channel estimation accuracy. In this paper, we propose to incorporate the nested array into analog array architecture by using RF switch networks with an objective of reducing the complexity and power consumption of the system. Based on the covariance fitting criterion, a joint Doppler and channel estimation method is proposed without need of discretizing the angle space, and thus the model mismatch effect can be totally eliminated. We also present an algorithmic implementation by solving the dual problem of the original one in order to reduce the computational complexity. Numerical simulations are provided to demonstrate the effectiveness and superiority of our proposed method.

cs.IT↗

arXiv2018

Speech Enhancement in Adverse Environments Based on Non-stationary Noise-driven Spectral Subtraction and SNR-dependent Phase Compensation

A two-step enhancement method based on spectral subtraction and phase spectrum compensation is presented in this paper for noisy speeches in adverse environments involving non-stationary noise and medium to low levels of SNR. The magnitude of the noisy speech spectrum is modified in the first step of the proposed method by a spectral subtraction approach, where a new noise estimation method based on the low frequency information of the noisy speech is introduced. We argue that this method of noise estimation is capable of estimating the non-stationary noise accurately. The phase spectrum of the noisy speech is modified in the second step consisting of phase spectrum compensation, where an SNR-dependent approach is incorporated to determine the amount of compensation to be imposed on the phase spectrum. A modified complex spectrum is obtained by aggregating the magnitude from the spectral subtraction step and modified phase spectrum from the phase compensation step, which is found to be a better representation of enhanced speech spectrum. Speech files available in the NOIZEUS database are used to carry extensive simulations for evaluation of the proposed method.

eess.AS↗

arXiv2018

Enhancement of Noisy Speech Exploiting an Exponential Model Based Threshold and a Custom Thresholding Function in Perceptual Wavelet Packet Domain

For enhancement of noisy speech, a method of threshold determination based on modeling of Teager energy (TE) operated perceptual wavelet packet (PWP) coefficients of the noisy speech by exponential distribution is presented. A custom thresholding function based on the combination of mu-law and semisoft thresholding functions is designed and exploited to apply the statistically derived threshold upon the PWP coefficients. The effectiveness of the proposed method is evaluated for car and multi-talker babble noise corrupted speech signals through performing extensive simulations using the NOIZEUS database. The proposed method outperforms some of the state-of-the-art speech enhancement methods both at high and low levels of SNRs in terms of the standard objective measures and the subjective evaluations including formal listening tests.

eess.AS↗

arXiv2018

Enhancement of Noisy Speech with Low Speech Distortion Based on Probabilistic Geometric Spectral Subtraction

A speech enhancement method based on probabilistic geometric approach to spectral subtraction (PGA) performed on short time magnitude spectrum is presented in this paper. A confidence parameter of noise estimation is introduced in the gain function of the proposed method to prevent subtraction of the overestimated and underestimated noise, which not only removes the noise efficiently but also prevents the speech distortion. The noise compensated magnitude spectrum is then recombined with the unchanged phase spectrum to produce a modified complex spectrum prior to synthesize an enhanced frame. Extensive simulations are carried out using the speech files available in the NOIZEUS database in order to evaluate the performance of the proposed method.

eess.AS↗

arXiv2018

Modeling of Teager Energy Operated Perceptual Wavelet Packet Coefficients with an Erlang-2 PDF for Real Time Enhancement of Noisy Speech

In this paper, for real time enhancement of noisy speech, a method of threshold determination based on modeling of Teager energy (TE) operated perceptual wavelet packet (PWP) coefficients of the noisy speech and noise by an Erlang-2 PDF is presented. The proposed method is computationally much faster than the existing wavelet packet based thresholding methods. A custom thresholding function based on a combination of mu-law and semisoft thresholding functions is designed and exploited to apply the statistically derived threshold upon the PWP coefficients. The proposed custom thresholding function works as a mu-law or a semisoft thresholding function or their combination based on the probability of speech presence and absence in a subband of the PWP transformed noisy speech. By using the speech files available in NOIZEUS database, a number of simulations are performed to evaluate the performance of the proposed method for speech signals in the presence of Gaussian white and street noises. The proposed method outperforms some of the state-of-the-art speech enhancement methods both at high and low levels of SNRs in terms of standard objective measures and subjective evaluations including formal listening tests.

eess.AS↗

arXiv2018

A Divide and Conquer Strategy for Musical Noise-free Speech Enhancement in Adverse Environments

A divide and conquer strategy for enhancement of noisy speeches in adverse environments involving lower levels of SNR is presented in this paper, where the total system of speech enhancement is divided into two separate steps. The first step is based on noise compensation on short time magnitude and the second step is based on phase compensation. The magnitude spectrum is compensated based on a modified spectral subtraction method where the cross-terms containing spectra of noise and clean speech are taken into consideration, which are neglected in the traditional spectral subtraction methods. By employing the modified magnitude and unchanged phase, a procedure is formulated to compensate the overestimation or underestimation of noise by phase compensation method based on the probability of speech presence. A modified complex spectrum based on these two steps are obtained to synthesize a musical noise free enhanced speech. Extensive simulations are carried out using the speech files available in the NOIZEUS database in order to evaluate the performance of the proposed method. It is shown in terms of the objective measures, spectrogram analysis and formal subjective listening tests that the proposed method consistently outperforms some of the state-of-the-art methods of speech enhancement for noisy speech corrupted by street or babble noise at very low as well as medium levels of SNR.

eess.AS↗

arXiv2017

A High-resolution DOA Estimation Method with a Family of Nonconvex Penalties

The low-rank matrix reconstruction (LRMR) approach is widely used in direction-of-arrival (DOA) estimation. As the rank norm penalty in an LRMR is NP-hard to compute, the nuclear norm (or the trace norm for a positive semidefinite (PSD) matrix) has been often employed as a convex relaxation of the rank norm. However, solving a nuclear norm convex problem may lead to a suboptimal solution of the original rank norm problem. In this paper, we propose to apply a family of nonconvex penalties on the singular values of the covariance matrix as the sparsity metrics to approximate the rank norm. In particular, we formulate a nonconvex minimization problem and solve it by using a locally convergent iterative reweighted strategy in order to enhance the sparsity and resolution. The problem in each iteration is convex and hence can be solved by using the optimization toolbox. Convergence analysis shows that the new method is able to obtain a suboptimal solution. The connection between the proposed method and the sparse signal reconstruction (SSR) is explored showing that our method can be regarded as a sparsity-based method with the number of sampling grids approaching infinity. Two feasible implementation algorithms that are based on solving a duality problem and deducing a closed-form solution of the simplified problem are also provided for the convex problem at each iteration to expedite the convergence. Extensive simulation studies are conducted to show the superiority of the proposed methods.

cs.IT↗

arXiv2015

Security-Reliability Trade-off Analysis of Multi-Relay Aided Decode-and-Forward Cooperation Systems

We consider a cooperative wireless network comprised of a source, a destination and multiple relays operating in the presence of an eavesdropper, which attempts to tap the source-destination transmission. We propose multi-relay selection scheme for protecting the source against eavesdropping. More specifically, multi-relay selection allows multiple relays to simultaneously forward the source's transmission to the destination, differing from the conventional single-relay selection where only the best relay is chosen to assist the transmission from the source to destination. For the purpose of comparison, we consider the classic direct transmission and single-relay selection as benchmark schemes. We derive closed-form expressions of the intercept probability and outage probability for the direct transmission as well as for the single-relay and multi-relay selection schemes over Rayleigh fading channels. It is demonstrated that as the outage requirement is relaxed, the intercept performance of the three schemes improves and vice versa, implying that there is a \emph{security versus reliability trade-off} (SRT). We also show that both the single-relay and multi-relay selection schemes outperform the direct transmission in terms of SRT, demonstrating the advantage of the relay selection schemes for protecting the source's transmission against the eavesdropping attacks. Finally, upon increasing the number of relays, the SRTs of both the single-relay and multi-relay selection schemes improve significantly and as expected, multi-relay selection outperforms single-relay selection.

cs.IT↗

arXiv2014

Relay-Selection Improves the Security-Reliability Trade-off in Cognitive Radio Systems

We consider a cognitive radio (CR) network consisting of a secondary transmitter (ST), a secondary destination (SD) and multiple secondary relays (SRs) in the presence of an eavesdropper. We rely on careful relay selection for protecting the ST-SD transmission against the eavesdropper with the aid of both single-relay and multi-relay selection. To be specific, only the "best" SR is chosen in the single-relay selection for assisting the secondary transmission, whereas the multi-relay selection invokes multiple SRs for simultaneously forwarding the ST's transmission to the SD. We analyze both the intercept probability and outage probability of the proposed single-relay and multi-relay selection schemes for the secondary transmission relying on realistic spectrum sensing. We also evaluate the performance of classic direct transmission and artificial noise based methods for the purpose of comparison with the proposed relay selection schemes. It is shown that as the intercept probability requirement is relaxed, the outage performance of the direct transmission, the artificial noise based and the relay selection schemes improves, and vice versa. This implies a trade-off between the security and reliability of the secondary transmission in the presence of eavesdropping attacks, which is referred to as the security-reliability trade-off (SRT). Furthermore, we demonstrate that the SRTs of the single-relay and multi-relay selection schemes are generally better than that of classic direct transmission. Moreover, as the number of SRs increases, the SRTs of the proposed single-relay and multi-relay selection approaches significantly improve. Finally, our numerical results show that as expected, the multi-relay selection scheme achieves a better SRT performance than the single-relay selection.

cs.IT↗