基于多元Laplace语音模型的语音增强算法

传统的短时谱估计语音增强算法通常假设语音谱分量相互独立,没有考虑语音谱分量间的相关性。针对这一问题,该文提出一种新的基于多元Laplace分布模型的短时谱估计算法。首先,假设语音的离散余弦变换(DCT)系数服从多元Laplace分布,以此利用谱分量间的相关性;在此基础上,利用多元随机矢量的高斯尺度混合模型表示,推导得到语音DCT系数矢量的最小均方误差(MMSE)估计的解析表达式;并进一步推导了基于该分布模型的语音存在概率,对最小均方误差估计子进行修正。实验结果表明,该算法在抑制背景噪声和减少语音失真等方面优于传统的语音增强方法。     

一种改进型MMSE语音增强方法

本文提出了一种改进型语音短时谱最小均方误差(MMSE)估计的增强方法。通过在每一帧及帧内每一频点对无音的概率(SAP)进行估计,得到Ephraim和MalahMMSE估计算法的改进形式。对增强后的语音客观和主观测试表明:在低信噪比条件下,相对于传统的谱减法和MMSE估计方法,这种改进的方法能更好的抑制背景噪声和残留的“音乐噪声”。     

Improved Frequency Domain Channel Estimation and Equalization for MIMO Wireless Communications

This paper introduces an improved frequency domain channel estimation method based on interpolation vectors for single carrier frequency domain equalization (SC-FDE) with the multiple-input multiple-output (MIMO) scheme. The proposed algorithm is derived by employing the least squares (LS) criterion, and a specified application for the wide sense stationary uncorrelated scattering (WSSUS) Rayleigh fading channel is presented. The channel frequency domain responses estimated at two adjacent pilot blocks are used to track the time-variant channel information, which can effectively improve the accuracy of channel estimation without significantly increasing complexity. Maximum mean square error (MMSE) frequency domain equalization based on the estimated channel is employed in the receiver to recover transmitted signals. This paper also investigates a training sequence design method for multiple transmit antennas and a noise variance estimation method. Numerical simulation results show that the proposed methods can perform very well for fading channels with long multipath delay and high Doppler spread.     

基于MMSE 信道估计的VRVP-MQAM 方法性能研究

信道估计对于VRVP-MQAM系统的整体性能至关重要,论述了VRVP-MQAM方法在实际应用中遇到的问题,针对假定CSI已知的传统研究方法,该文通过采用MMSE算法进行信道估计,研究了MMSE估计误差对VRVP-MQAM方法的ASE性能影响,并在Rayleigh衰落信道下进行了仿真,仿真结果表明:与信道状态已知(=1)相比,MMSE估计误差(=0.9时)会产生1~3 dB左右的ASE性能差距;当平均SNR为20 dB时,MMSE算法下信道状态的平均频谱效率为0.3 bps/Hz,比理想信道状态的平均频谱效率低。因此,VRVP-MQAM方法的应用将会越来越广泛。     

一种基于短时谱估计和人耳掩蔽效应的语音增强算法

该文结合短时谱估计算法和人耳掩蔽效应提出了一种单通道语音增强算法。该算法在MMSE准则下采用了非固定参数的语音跟踪,并且引入人耳掩蔽效应动态的确定增强滤波器的传递函数以适应语音信号的变化。实验结果表明:该算法使降噪后的语音信号有较小的语音失真并且很好地抑制了音乐噪声。     

采用模型自适应的语音转换方法

针对非对称语音库情况下的语音转换,提出了一种有效的基于模型自适应的语音转换方法。首先,通过最大后验概率（Maximum A Posteriori,MAP）方法从背景模型分别自适应训练得到源说话人和目标说话人的模型;然后,通过说话人模型中的均值向量训练得到频谱特征的转换函数;并进一步与传统的INCA转换方法相结合,提出了基于模型自适应的INCA语音转换方法,有效实现了源说话人频谱特征向目标说话人频谱特征的转换。通过客观测试和主观测听实验对提出的方法进行评价,实验结果表明,与INCA语音转换方法相比,本文提出的方法可以取得更低的倒谱失真、更高的语音感知质量和目标倾向度;同时更接近传统基于对称语音库的高斯混合模型（Gaussian Mixture Model,GMM）的语音转换方法的效果。      

Analysis of low-complexity windowed DFT-based MMSE channelestimator for OFDM systems

Low-complexity windowed discrete Fourier transform (DFT)-based minimum mean square error (MMSE) channel estimators are proposed and analyzed for both the interpolation and noninterpolation cases for orthogonal frequency-division multiplexing (OFDM) mobile communications systems. In the proposed method, the frequency domain data windowing is used to reduce the aliasing errors for the interpolation case and get better noise filtering performance for the noninterpolation case. The time domain MMSE weighting is also used to suppress the channel noise for both cases. Moreover, the optimal generalized Hanning window shape is searched to minimize the channel estimation mean square error (MSE). Analysis and simulation results show that the proposed method performance is close to the optimal MMSE estimator and is much better than the direct DFT-based estimator for both cases. Compared with the optimal MMSE estimator, however, the computation load of the proposed method can be significantly reduced because the IDFT/DFT transforms can be implemented with the fast algorithms IFFT/FFT     

EM Based Data Detection Algorithm for Downlink of a Single Hop Relaying Network

This work proposes an Expectation–Maximization (EM) based joint channel estimation and data detection algorithm and a time sharing hopped cooperative communication for the downlink of a communication network. The algorithm was applied to the downlink of a single hop relaying network. The relaying technique is a variant of the decode and forward technique. However, it is a time sharing single hop cooperative communication technique applied to a multi-carrier code division multiple access used in cellular networks. The proposed EM algorithm is compared with the minimum mean square estimation (MMSE) method using simulation. In addition, the proposed relaying technique is compared with cooperative communication on the considered cellular network for frequency-selective fading channels. Simulation results illustrate that our proposed EM algorithm performs better than the MMSE algorithm and that our proposed relaying technique is better than the normal case and the cooperative diversity case.     

An asynchronous multiuser CDMA detector based on the Kalman filter

We introduce a multiuser receiver based on the Kalman filter, which can be used for joint symbol detection and channel estimation. The proposed algorithm has the advantage of working even when the spreading codes used have a period larger than one symbol interval (“long codes”), unlike adaptive equalizer-type detectors. Simulation results which demonstrate the performance advantage of the proposed receiver over the conventional detector, the minimum mean squared error (MMSE) detector and a recursive least squares (RLS) multiuser detector are presented. A thorough comparison of the MMSE detector and the proposed detector is attempted because the Kalman filter also solves the MMSE parameter estimation problem, and it is concluded that, because the state space model assumed by the Kalman filter fits the code division multiple access (CDMA) system exactly, a multiuser detector based on the Kalman filter must necessarily perform better than a nonrecursive, finite-length MMSE detector. The computational complexity of the detector and its use in channel estimation are also studied     

Linear expansions for frequency selective channels in OFDM

Modeling the frequency selective fading channels as random processes, we employ a linear expansion based on the Karhunen–Loeve (KL) series representation involving a complete set of orthogonal deterministic vectors with a corresponding uncorrelated random coefficients. Focusing on OFDM transmissions through frequency selective fading, this paper pursues a computationally efficient, pilot-aided linear minimum mean square error (MMSE) uncorrelated KL series expansion coefficients estimation algorithm. Based on such an expansion, no matrix inversion is required in the proposed MMSE estimator. Moreover, truncation in the linear expansion of channel is achieved by exploiting the optimal truncation property of the KL expansion resulting in a smaller computational load on the estimation algorithm. The performance of the proposed approach is studied through analytical and experimental results. We first exploit the performance of the MMSE channel estimator based on the evaluation of minimum Bayesian MSE. We also provide performance analysis results studying the influence of the effect of SNR and correlation mismatch on the estimator performance. Simulation results confirm our theoretical results and illustrate that the proposed algorithm is capable of tracking fast fading and improving performance.     

Hybrid SNR-Adaptive Multiuser Detectors for SDMA-OFDM Systems

Multiuser detection (MUD) and channel estimation techniques in space-division multiple-access aided orthogonal frequency-division multiplexing systems recently has received intensive interest in receiver design technologies. The maximum likelihood (ML) MUD that provides optimal performance has the cost of a dramatically increased computational complexity. The minimum mean-squared error (MMSE) MUD exhibits poor performance, although it achieves lower computational complexity. With almost the same complexity, an MMSE with successive interference cancellation (SIC) scheme achieves a better bit error rate performance than a linear MMSE multiuser detector. In this paper, hybrid ML-MMSE with SIC adaptive multiuser detection based on the joint channel estimation method is suggested for signal detection. The simulation results show that the proposed method achieves good performance close to the optimal ML performance at low SNR values and a low computational complexity at high SNR values.     

MC-CDMA系统降维盲空时信道估计及信号检测

该文提出了适用于频率选择性瑞利衰落信道中的MC-CDMA系统降维盲空时信道估计及信号检测算法。对接收数据矢量进行解相关滤波,得到对应于每个用户的无多址干扰(MAI)的降维信号,由白化滤波后的降维数据组成的矩阵的秩1近似来进行盲信道估计。将白化滤波后的降维数据矢量看作是单用户的系统,研究了迭代最小二乘(ILS)检测和最小均方误差(MMSE)检测。仿真结果验证了该文算法的有效性。     

无线传感器网络SL-n迭代定位算法

无线传感器网络迭代多边定位算法在迭代定位过程中使用全局最小均方估计(MMSE)方法估计盲节点的位置,导致算法定位误差很大,缺乏稳健性。针对此问题,提出了SL-n估计方法,该方法首先把某盲节点的所有参考节点进行分组,通过三边法或局部MMSE方法求出的每组相应样本值,并用这些样本值估计出此盲节点的位置。仿真实验表明,当部分参考节点的参考误差较大时,该方法优于全局MMSE处理方法,可以有效降低定位误差。     

基于实值离散Gabor变换的联合时频域语音增强

传统变换域语音增强方法对语音做短时平稳性假设,这会造成对语音信号和噪声信号谱估计不准确,从而导致语音失真和残留噪声。本文提出一种从联合时频域进行语音增强的方法,该算法无需对语音做短时平稳假设。算法采用具有最佳能量聚集特性的高斯变换核函数,利用能快速实现的实值离散Gabor变换（RDGT）将语音信号变换到联合时频域,然后利用语音和噪声谱服从高斯分布的假设和无语音概率的思想进行基于最小均方误差的语音对数谱估计,采用改进的最小受控递归平均算法（IMCRA）进行噪声时频谱估计,在得到纯净语音的谱估计后利用实值离散Gabor逆变换获得纯净语音估计。实验表明,该算法相比频域变换算法具有较好的语音去噪度和较低的语音失真度。      

Training sequence based channel estimation for indoor visible light communication system

Channel estimation is a key technology in indoor wireless visible light communications(VLCs).Using the training sequence(TS),this paper investigates the channel estimation in indoor wireless visible light communications.Based on the propagation and signal modulation characteristics of visible light,a link model for the indoor wireless visible light communications is established.Using the model,three channel estimation methods,i.e.,the correlation method,the least square(LS) method and the minimum mean square error(MMSE) method,are proposed.Moreover,the performances of the proposed three methods are evaluated by computer simulation.The results show that the performance of the correlation method is the worst,the LS method is suitable for higher signal to noise ratio(SNR),and the MMSE method obtains the best performance at the expense of highest complexity.     

An Adaptive Multiuser Detection Algorithm for CDMA Systems Using Antenna Diversity

Based on the minimum mean squared error (MMSE) between the data stream and the linear combiner output, a new multiuser detection (MUD) algorithm that combines space–time (ST) processing and antenna array on direct-sequence CDMA signals is proposed. The proposed ST-MUD algorithm is proved to be equivalent to two existing MMSE-based ST-MUD algorithms, and the theoretical BER performances for all the three algorithms are the same. The most attractive feature of the new ST-MUD algorithm is based on the fact that the new method does not require explicit estimation of channel and signaling information. This avoids any channel estimation error, and the method is thus more robust and more accurate than the other two ST-MUD algorithms in practical implementation. Adaptation of the proposed ST-MUD algorithm is implemented by using training sequences. Performance of this new multiuser detector is compared with that of two existing MMSE multiuser detectors and the conventional single-user space–time rake receiver through simulations. The proposed ST-MUD algorithm provides a performance better than existing algorithms and is especially suitable for practical CDMA systems.     

Genetic Grey Wolf Optimizer Based Channel Estimation in Wireless Communication System

Various methods are available for channel estimation in the orthogonal frequency division multiplexing and orthogonal frequency and code division multiplexing (OFCDM) based wireless communication schemes. Along with this, the most utilized techniques are namely the minimum mean square error (MMSE) and least square (LS). The process of LS channel estimation method is simple but it occupies a very high mean square error. On the other hand, the performance of MMSE is better than LS in terms of SNR, though it shows high computational complexity. Compared to MMSE and LS based techniques, the combination of MMSE and LS techniques using evolutionary programming reduces the error significantly to receive exact signal. In this study, we propose a hybrid method namely GGWO that includes grey wolf optimization (GWO) and genetic algorithms (GA) for estimate the channel in MIMO–OFCDM schemes. At first, the best channel is estimated using GWO and afterwards, the MMSE and LS are hybridized through GA for calculating the best channel to decrease error. Overall, the GWO and GA contribute in fine tuning the obtained channel scheme so that the channel model is derived further to correlate with the ideal scheme. Our results demonstrate that the proposed scheme is superior to conventional MMSE and LS in terms of BER and SNR.     

Improved ML Channel Estimation for Uplink MC-CDMA Systems in Closely Spaced Multipath Channels

In order to attain near-single user performance in uplink multicarrier code- division multiple-access (MC-CDMA) systems, multiuser detection (MUD) methods may be employed which rely on simultaneous estimation of the channel frequency responses of multiple users. Pilot symbol assisted (PSA) channel estimation is needed in fast fading channels and it can be performed either by applying maximum likelihood (ML) criterion or minimum mean-squared error (MMSE) criterion. The performance of ML estimation technique degrades significantly in the case of fractionally spaced (FS) multipath channels where dominant paths are closely spaced with respect to the time resolution of the system. In such situation, the number of effective paths (which contribute more towards signal power) becomes considerably less than the actual number of multipaths at low and moderate SNR values. We propose an improved ML estimation method which considers only effective paths during the estimation process. The proposed method performs nearly identical to the MMSE estimation method and it can also provide significant reduction in the computational complexity when a large number of users are accommodated in the system.     

Minimum mean squared error impulse noise estimation andcancellation

Impulses are infrequent bursts of high amplitude noise. A wide-band communications or data acquisition receiver has a fast sampling rate, so it can capture many samples of each impulse waveform. The arrival of an impulse can be identified by its distinct waveform and amplitude. The paper models impulse waveforms as a vector subspace of low dimension. Formulas are derived for the minimum mean squared error (MMSE) estimation of the arrival time and amplitudes of impulses, given that a set of vectors that spans the subspace is known. Formulas are also derived for the adaptive MMSE estimation of a set of vectors that spans the subspace. The values of the mean squared error (MSE) of the amplitude estimates are determined. It is shown how the theory can be used to cancel impulse noise. Correlated impulse noise arriving at a reference input can be used to estimate and cancel the primary input impulse noise. The MMSE coefficients for impulse noise cancellation are derived and presented. Simulations are presented that use the equations and methods derived in the paper for modeling and canceling impulse noise measured on copper telephone loops for asymmetric digital subscriber lines (ADSL)     

Minimum mean-square error frequency-domain equalisation in unique-word based single-carrier systems

Optimal and sub-optimal linear minimum mean-square error (MMSE) frequency-domain equalisers are derived for single-carrier systems employing unique-word based block transmissions. These equalisers differ from those presented in the literature for cyclic-prefix based systems. The effects of channel estimation on the performance of these equalisers are also addressed. It is shown that the proposed MMSE equaliser performs 2 dB better than the conventional MMSE equaliser.