Takehiro KITAMURA Mahfuzul ISLAM Takashi HISAKADO Osami WADA
High-speed flash ADCs are useful in high-speed applications such as communication receivers. Due to offset voltage variation in the sub-micron processes, the power consumption and the area increase significantly to suppress variation. As an alternative to suppressing the variation, we have developed a flash ADC architecture that selects the comparators based on offset voltage ranking for reference generation. Specifically, with the order statistics as a basis, our method selects the minimum number of comparators to obtain equally spaced reference values. Because the proposed ADC utilizes offset voltages as references, no resistor ladder is required. We also developed a time-domain sorting mechanism for the offset voltages to achieve on-chip comparator selection. We first perform a detailed analysis of the order statistics based selection method and then design a 4-bit ADC in a commercial 65-nm process and perform transistor-level simulation. When using 127 comparators, INLs of 20 virtual chips are in the range of -0.34LSB/+0.29LSB to -0.83LSB/+0.74LSB, and DNLs are in the range of -0.33LSB/+0.24LSB to -0.77LSB/+1.18LSB at 1-GS/s operation. Our ADC achieves the SNDR of 20.9dB at Nyquist-frequency input and the power consumption of 0.84mW.
Quan TIAN Tianshuang QIU Jitong MA Jingchun LI Rong LI
In array signal processing, many methods of handling cases of impulsive noise with an alpha-stable distribution have been studied. By introducing correntropy with a robust statistical property, this paper proposes a novel fractional lower order correntropy (FLOCR) method. The FLOCR-based estimator for array outputs is defined and applied with multiple signal classification (MUSIC) to estimate the direction of arrival (DOA) in alpha-stable distributed noise environments. Comprehensive Monte Carlo simulation results demonstrate that FLOCR-MUSIC outperforms existing algorithms in terms of root mean square error (RMSE) and the probability of resolution, especially in the presence of highly impulsive noise.
Naoto SASAOKA Eiji AKAMATSU Arata KAWAMURA Noboru HAYASAKA Yoshio ITOH
Speech enhancement has been proposed to reduce the impulsive noise whose frequency characteristic is wideband. On the other hand, it is challenging to reduce the ringing sound, which is narrowband in impulsive noise. Therefore, we propose the modeling of the ringing sound and its estimation by a linear predictor (LP). However, it is difficult to estimate the ringing sound only in noisy speech due to the auto-correlation property of speech. The proposed system adopts the 4th order moment-based adaptive algorithm by noticing the difference between the 4th order statistics of speech and impulsive noise. The brief analysis and simulation results show that the proposed system has the potential to reduce ringing sound while keeping the quality of enhanced speech.
We propose a deep learning-based model for classifying pathological voices using a convolutional neural network and a feedforward neural network. The model uses combinations of heterogeneous parameters, including mel-frequency cepstral coefficients, linear predictive cepstral coefficients and higher-order statistics. We validate the accuracy of this model using the Massachusetts Eye and Ear Infirmary (MEEI) voice disorder database and the Saarbruecken Voice Database (SVD). Our model achieved an accuracy of 99.3% for MEEI and 75.18% for SVD. This model achieved an accuracy that is 7.18% higher than that of competitive models in previous studies.
Shogo KOYANAGI Teruyuki MIYAJIMA
In this paper, we consider full-duplex (FD) relay networks with filter-and-forward (FF)-based multiple relays (FD-FF), where relay filters jointly mitigate self-interference (SI), inter-relay interference (IRI), and inter-symbol interference. We consider the filter design problem based on signal-to-noise-plus-interference ratio maximization subject to a total relay transmit power constraint. To make the problem tractable, we propose two methods: one that imposes an additional constraint whereby the filter responses to SI and IRI are nulled, and the other that makes i.i.d. assumptions on the relay transmit signals. Simulation results show that the proposed FD-FF scheme outperforms a conventional FF scheme in half-duplex mode. We also consider the filter design when only second-order statistics of channel path gains are available.
Satoshi NAGAI Teruyuki MIYAJIMA
In this paper, we consider filter-and-forward relay beamforming using orthogonal frequency-division multiplexing (OFDM) in the presence of inter-block interference (IBI). We propose a filter design method based on a constrained max-min problem, which aims to suppress IBI and also avoid deep nulls in the frequency domain. It is shown that IBI can be suppressed completely owing to the employment of beamforming with multiple relays or multiple receive antennas at each relay when perfect channel state information (CSI) is available. In addition, we modify the proposed method to cover the case where only the partial CSI for relay-receiver channels is available. Numerical simulation results show that the proposed method significantly improves the performance as the number of relays and antennas increases due to spatial diversity, and the modified method can make use of the channel correlation to improve the performance.
Yuya SUGIMOTO Shigeki MIYABE Takeshi YAMADA Shoji MAKINO Biing-Hwang JUANG
MUltiple SIgnal Classification (MUSIC) is a standard technique for direction of arrival (DOA) estimation with high resolution. However, MUSIC cannot estimate DOAs accurately in the case of underdetermined conditions, where the number of sources exceeds the number of microphones. To overcome this drawback, an extension of MUSIC using cumulants called 2q-MUSIC has been proposed, but this method greatly suffers from the variance of the statistics, given as the temporal mean of the observation process, and requires long observation. In this paper, we propose a new approach for extending MUSIC that exploits higher-order moments of the signal for the underdetermined DOA estimation with smaller variance. We propose an estimation algorithm that nonlinearly maps the observed signal onto a space with expanded dimensionality and conducts MUSIC-based correlation analysis in the expanded space. Since the dimensionality of the noise subspace is increased by the mapping, the proposed method enables the estimation of DOAs in the case of underdetermined conditions. Furthermore, we describe the class of mapping that allows us to analyze the higher-order moments of the observed signal in the original space. We compare 2q-MUSIC and the proposed method through an experiment assuming that the true number of sources is known as prior information to evaluate in terms of the bias-variance tradeoff of the statistics and computational complexity. The results clarify that the proposed method has advantages for both computational complexity and estimation accuracy in short-time analysis, i.e., the time duration of the analyzed data is short.
Mingda WANG Gaolei FEI Guangmin HU
Flow classification is of great significance for network management. Machine-learning-based flow classification is widely used nowadays, but features which depict the non-Gaussian characteristics of network flows are still absent. In this paper, we propose the Windowed Higher-order Statistical Analysis (WHOSA) for machine-learning-based flow classification. In our methodology, a network flow is modeled as three different time series: the flow rate sequence, the packet length sequence and the inter-arrival time sequence. For each sequence, both the higher-order moments and the largest singular values of the Bispectrum are computed as features. Some lower-order statistics are also computed from the distribution to build up the feature set for contrast, and C4.5 decision tree is chosen as the classifier. The results of the experiment reveals the capability of WHOSA in flow classification. Besides, when the classifier gets fully learned, the WHOSA feature set exhibit stronger discriminative power than the lower-order statistical feature set does.
Da Sol KIM Taek Lyul SONG Darko MUŠICKI
In this paper, we propose a new data association method termed the highest probability data association (HPDA) and apply it to real-time recursive nonlinear tracking in heavy clutter. The proposed method combines the probabilistic nearest neighbor (PNN) with a modified probabilistic strongest neighbor (PSN) approach. The modified PSN approach uses only the rank of the measurement amplitudes. This approach is robust as exact shape of amplitude probability density function is not used. In this paper, the HPDA is combined with particle filtering for nonlinear target tracking in clutter. The measurement with the highest measurement-to-track data association probability is selected for track update. The HPDA provides the track quality information which can be used in for the false track termination and the true track confirmation. It can be easily extended to multi-target tracking with nonlinear particle filtering. The simulation studies demonstrate the HPDA functionality in a hostile environment with high clutter density and low target detection probability.
Ryoichi MIYAZAKI Hiroshi SARUWATARI Kiyohiro SHIKANO
We propose a structure-generalized blind spatial subtraction array (BSSA), and the theoretical analysis of the amounts of musical noise and speech distortion. The structure of BSSA should be selected according to the application, i.e., a channelwise BSSA is recommended for listening but a conventional BSSA is suitable for speech recognition.
David COURNAPEAU Tatsuya KAWAHARA
A new online, unsupervised voice activity detection (VAD) method is proposed. The method is based on a feature derived from high-order statistics (HOS), enhanced by a second metric based on normalized autocorrelation peaks to improve its robustness to non-Gaussian noises. This feature is also oriented for discriminating between close-talk and far-field speech, thus providing a VAD method in the context of human-to-human interaction independent of the energy level. The classification is done by an online variation of the Expectation-Maximization (EM) algorithm, to track and adapt to noise variations in the speech signal. Performance of the proposed method is evaluated on an in-house data and on CENSREC-1-C, a publicly available database used for VAD in the context of automatic speech recognition (ASR). On both test sets, the proposed method outperforms a simple energy-based algorithm and is shown to be more robust against the change in speech sparsity, SNR variability and the noise type.
Ji-Yeoun LEE Sangbae JEONG Hong-Shik CHOI Minsoo HAHN
This work proposes new features to improve the pathological voice quality classification performance. They are the means, the variances, and the perturbations of the higher-order statistics (HOS) such as the skewness and the kurtosis. The HOS-based features show meaningful differences among normal, grade 1, grade 2, and grade 3 voices classified in the GRBAS scale. The jitter, the shimmer, the harmonic-to-noise ratio (HNR), and the variance of the short-time energy are utilized as the conventional features. The performances are measured by the classification and regression tree (CART) method. Specifically, the CART-based method by utilizing both the conventional features and the HOS-based ones shows its effectiveness in the pathological voice quality measurement, with the classification accuracy of 87.8%.
Ki-Hong PARK Hong-Chuan YANG Young-Chai KO
Transmit diversity systems based on orthogonal space-time block coding (OSTBC) usually suffer from rate loss and power spreading. Proper antenna selection scheme can help to more effectively utilize the transmit antennas and transmission power in such systems. In this paper, we propose a new antenna selection scheme for such systems based on the idea of antenna switching. In particular, targeting at reducing the number of pilot channels and RF chains, the transmitter now replaces the antennas with the lowest received SNR with unused ones if the output SNR of space time decoder at the receiver is below a certain threshold. With this new scheme, not only the number of pilot channels and RF chains to be implemented is decreased, the average amount of feedback information is also reduced. To analyze the performance of this scheme, we derive the exact integral closed form for the probability density function (PDF) of the received SNR. We show through numerical examples that the proposed scheme offers better performance than traditional OSTBC systems using all available transmitting antennas, with a small amount of feedback information. We also examine the effect of different antenna configuration and feedback delay.
Ji-Yeoun LEE Sangbae JEONG Minsoo HAHN
Combination of mutually complementary features is necessary to cope with various changes in pattern classification between normal and pathological voices. This paper proposes a method to improve pathological/normal voice classification performance by combining heterogeneous features. Different combinations of auditory-based and higher-order features are investigated. Their performances are measured by Gaussian mixture models (GMMs), linear discriminant analysis (LDA), and a classification and regression tree (CART) method. The proposed classification method by using the CART analysis is shown to be an effective method for pathological voice detection, with a 92.7% classification performance rate. This is a noticeable improvement of 54.32% compared to the MFCC-based GMM algorithm in terms of error reduction.
Volodymyr PONOMARYOV Alberto ROSALES Francisco GALLEGOS Igor LOBODA
We present a novel algorithm that can suppress impulsive noise in video colour sequences. It uses order statistics, and directional and adaptive processing techniques.
Younghyun JEON Sungho JEON Sanghoon LEE
It is well known that the diversity gain attained by DCA (Dynamic Channel Allocation) is generally very high over OFDM (Orthogonal Frequency Division Multiplexing)-based broadband networks. This paper introduces a numerical approach for measuring the performance gain afforded by DCA. In the mathematical analysis, the property of order statistics is adopted to derive the upper bound of the expected throughput via the use of DCA. In the simulation, it was possible to achieve a gain of 5 dB by exploiting multi-user and spectral diversities when the number of users is 16 and the total number of subcarriers is 256.
Muhammad TUFAIL Masahide ABE Masayuki KAWAMATA
In this paper, we propose to employ an extension to the natural gradient algorithm for robust Independent Component Analysis against outliers. The standard natural gradient algorithm does not exhibit this property since it employs nonrobust sample estimates for computing higher order moments. In order to overcome this drawback, we propose to use robust alternatives to higher order moments, which are comparatively less sensitive to outliers in the observed data. Some computer simulations are presented to show that the proposed method, as compared to the standard natural gradient algorithm, gives better performance in the presence of outlying data.
Yong XIANG Wensheng YU Jingxin ZHANG Senjian AN
This paper presents a new method for blind source separation by exploiting phase and frequency redundancy of cyclostationary signals in a complementary way. It requires a weaker separation condition than those methods which only exploit the phase diversity or the frequency diversity of the source signals. The separation criterion is to diagonalize a polynomial matrix whose coefficient matrices consist of the correlation and cyclic correlation matrices, at time delay τ= 0, of multiple measurements. An algorithm is proposed to perform the blind source separation. Computer simulation results illustrate the performance of the new algorithm in comparison with the existing ones.
In this paper, a simple blind algorithm for a beamforming antenna is proposed. This algorithm exploits the property of cyclostationary signals whose cyclic autocorrelation function depends on delay as well as frequency. The cost function is the mean square error between the delay product of the beamformer output and a complex exponential. Exploiting the delay greatly reduces the possibility of capturing undesired signals. Through analysis of the minima of the non-quadratic cost function, conditions to extract a single signal are derived. Application of this algorithm to code-division multiple-access systems is considered, and it is shown through simulation that the desired signal can be extracted by appropriately choosing the delay as well as the frequency.
Mitsuru KAWAMOTO Yujiro INOUYE
The present paper deals with the blind deconvolution of a Multiple-Input Multiple-Output Finite Impulse Response (MIMO-FIR) system. To deal with the blind deconvolution problem using the second-order statistics (SOS) of the outputs, Hua and Tugnait considered it under the conditions that a) the FIR system is irreducible and b) the input signals are spatially uncorrelated and have distinct power spectra. In the present paper, the problem is considered under a weaker condition than the condition a). Namely, we assume that c) the FIR system is equalizable by means of the SOS of the outputs. Under b) and c), we show that the system can be blindly identified up to a permutation, a scaling, and a delay using the SOS of the outputs. Moreover, based on this identifiability, we show a novel necessary and sufficiently condition for solving the blind deconvolution problem, and then, based on the condition, we propose a new algorithm for finding an equalizer using the SOS of the outputs, while Hua and Tugnait have not proposed any algorithm for solving the blind deconvolution under the conditions a) and b).