This paper proposes a robust bilateral filter which can handle mixed Gaussian and impulsive noise by hybridizing the conventional bilateral filter and the switching median filter. The effectiveness of the proposed method is verified in comparison with other conventional methods by some experiments using the natural digital images.

This paper introduces our developed noise robust speech communication techniques and describes its implementation to a smart info-media system, i.e., a small robot. Our designed speech communication system consists of automatic speech detection, recognition, and rejection. By using automatic speech detection and recognition, an observed speech waveform can be recognized without a manual trigger. In addition, using speech rejection, this system only accepts registered speech phrases and rejects any other words. In other words, although an arbitrary input speech waveform can be fed into this system and recognized, the system responds only to the registered speech phrases. The developed noise robust speech processing can reduce various noises in many environments. In addition to the design of noise robust speech recognition, the LSI design of this system has been introduced. By using the design of speech recognition application specific IC (ASIC), we can simultaneously realize low power consumption and real-time processing. This paper describes the LSI architecture of this system and its performances in some field experiments. In terms of current speech recognition accuracy, the system can realize 85-99% under 0-20dB SNR and echo environments.

This paper presents a voice conversion (VC) technique for noisy environments, where parallel exemplars are introduced to encode the source speech signal and synthesize the target speech signal. The parallel exemplars (dictionary) consist of the source exemplars and target exemplars, having the same texts uttered by the source and target speakers. The input source signal is decomposed into the source exemplars, noise exemplars and their weights (activities). Then, by using the weights of the source exemplars, the converted signal is constructed from the target exemplars. We carried out speaker conversion tasks using clean speech data and noise-added speech data. The effectiveness of this method was confirmed by comparing its effectiveness with that of a conventional Gaussian Mixture Model (GMM)-based method.

Spectral subtraction (SS) is an additive noise removal method which is derived in an extensive framework. In spectral subtraction, it is assumed that speech and noise spectra follow Gaussian distributions and are independent with each other. Hence, noisy speech also follows a Gaussian distribution. Spectral subtraction formula is obtained by maximizing the likelihood of noisy speech distribution with respect to its variance. However, it is well known that noisy speech observed in real situations often follows a heavy-tailed distribution, not a Gaussian distribution. In this paper, we introduce a q-Gaussian distribution in the non-extensive statistics to represent the distribution of noisy speech and derive a new spectral subtraction method based on it. We found that the q-Gaussian distribution fits the noisy speech distribution better than the Gaussian distribution does. Our speech recognition experiments using the Aurora-2 database showed that the proposed method, q-spectral subtraction (q-SS), outperformed the conventional SS method.

I propose an acoustic model adaptation method using bases constructed through the sparse principal component analysis (SPCA) of acoustic models trained in a clean environment. I perform experiments on adaptation to a new speaker and noise. The SPCA-based method outperforms the PCA-based method in the presence of babble noise.

An instrumental variable (IV) based linear estimator is proposed for effective target localization in sensor network by using time-difference-of-arrival (TDOA) measurement. Although some linear estimation approaches have been proposed in much literature, the target localization based on TDOA measurement still has a room for improvement. Therefore, we analyze the estimation errors of existing localization estimators such as the well-known quadratic correction least squares (QCLS) and the robust least squares (RoLS), and demonstrate advantages of the proposition by comparing the estimation errors mathematically and showing localization results through simulation. In addition, a recursive form of the proposition is derived to consider a real time application.

To mitigate the inter-cell interference in multicell downlink systems, this letter consider the robust precoder design for multicell cooperation where the knowledge of channel state available at the base station is imperfect. Assuming that imperfect channel state information (CSI) can be exchanged among cells but with no data sharing, we investigate the worst-case performance optimization problem with bounded CSI error. Our objective is to minimize the weighted sum mean-square-error (MSE) subject to per-base-station power constraints. A distributed solution is obtained by reformulating the upper bound of MSE and exploiting the Lagrangian method for the optimal problem. Simulation results demonstrate that the proposed algorithm is robust to guarantee the worst-case sum rate performance and has lower computational complexity than the SINR-based design.

In this paper we show that self synchronous circuits can provide robust operation in both soft error prone and low voltage operating environments. Self synchronous circuits are shown to be self checking, where a soft error will either cause a detectable error or halt operation of the circuit. A watchdog circuit is proposed to autonomously detect dual-rail '11' errors and prevent propagation, with measurements in 65 nm CMOS showing seamless operation from 1.6 V to 0.37 V. Compared to a system without the watchdog circuit size and energy-per-operation is increased 6.9% and 16% respectively, while error tolerance to noise is improved 83% and 40% at 1.2 V and 0.4 V respectively. A circuit that uses the dual-pipeline circuit style as redundancy against permanent faults is also presented and 40 nm CMOS measurement results shows correct operation with throughput of 1.2 GHz and 810 MHz at 1.1 V before and after disabling a faulty pipeline stage respectively.

During the production of speech signals, the vowel onset point is an important event containing important information for many speech processing tasks, such as consonant-vowel unit recognition and speech end-points detection. In order to realize accurate automatic detection of vowel onset points, this paper proposes a reliable method using the energy characteristics of homomorphic filtered spectral peaks. The homomorphic filtering helps to separate the slowly varying vocal tract system characteristics from the rapidly fluctuating excitation characteristics in the cepstral domain. The distinct vocal tract shape related to vowels is obtained and the peaks in the estimated vocal tract spectrum provide accurate and stable information for VOP detection. Performance of the proposed method is compared with the existing method which uses the combination of evidence from the excitation source, spectral peaks, and modulation spectrum energies. The detection rate with different time resolutions, together with the missing rate and spurious rate, are used for comprehensive evaluation of the performance on continuous speech taken from the TIMIT database. The detection accuracy of the proposed method is 74.14% for ±10 ms resolution and it increases to 96.33% for ±40 ms resolution with 3.67% missing error and 4.14% spurious error, much better than the results obtained by the combined approach at each specified time resolution, especially the higher resolutions of ±10±30 ms. In the cases of speech corrupted by white noise, pink noise and f-16 noise, the proposed method also shows significant improvement in the performance compared with the existing method.

In this paper, a robust cyclic ADC architecture with β-encoder is proposed and circuit scheme using switched-capacitor (SC) circuit is introduced. Different from the conventional binary ADC, the redundancy of proposed cyclic ADC outputs β-expansion code and has an advantage of error correction. This feature makes ADC robust against the offset of comparator capacitor mismatch and finite DC gain of amplifier in multiplying-DAC (MDAC). Because the power penalty of high-gain wideband amplifier and the required accuracy of circuit elements for high resolution ADC can be relaxed, the proposed architecture is suitable for deep submicron CMOS technologies beyond 90 nm. We also propose a β-value estimation algorithm to realize high accuracy ADC based on β-expansion. The simulation results show the effectiveness of proposed architecture and robustness of β-encoder.

In this letter, the problem of feature quantization in robust hashing is studied from the perspective of approximate nearest neighbor (ANN). We model the features of perceptually identical media as ANNs in the feature set and show that ANN indexing can well meet the robustness and discrimination requirements of feature quantization. A feature quantization algorithm is then developed by exploiting the random-projection based ANN indexing. For performance study, the distortion tolerance and randomness of the quantizer are analytically derived. Experimental results demonstrate that the proposed work is superior to state-of-the-art quantizers, and its random nature can provide robust hashing with security against hash forgery.

If a duty ratio, a load resistance and an input voltage in a boost DC-DC converter are changed, the dynamic characteristics are varied greatly, that is, the boost DC-DC converter has non-linear characteristics. In many applications of the boost DC-DC converters, the loads cannot be specified in advance, and they will be changed suddenly from no load to full load. In the boost DC-DC converter, a conventional single controller cannot be adapted to change dynamics and it occurs large output voltage variation. In this paper, an approximate 2-degree-of-freedom (A2DOF) digital controller for suppressing the change of step response characteristics and the variation of an output voltage in load sudden change is proposed. Experimental studies using a micro-processor for the controller demonstrate that this type of digital controller is effective to suppress the variations of the output voltages.

Robustness is one of the significant properties in wireless sensor networks because sensor nodes and wireless links are subjected to frequent failures. Once these failures occur, system performance falls into critical condition due to increases in traffic and losses of connectivity and reachability. Most of the existing studies on sensor networks, however, do not conduct quantitative evaluation on robustness and do not discuss what brings in robustness. In this paper, we define and evaluate robustness of wireless sensor networks and show how to improve them. By computer simulation, we show that receiver-initiated MAC protocols are more robust than sender-initiated ones and a simple detour-routing algorithm has more than tripled robustness than the simple minimum-hop routing algorithm.

We consider wireless secure communications between a source and a destination aided by a multi-antenna relay, in the presence of an eavesdropper. In particular, two cooperation schemes of the relay are explored: cooperative relaying (CR) and cooperative jamming (CJ). We first investigate the transmit weight optimization of CR and CJ, for both cases with and without the eavesdropper's channel state information (ECSI). Then, for the case with ECSI, we derive the conditions under which CR achieves a higher secrecy rate than CJ; for the case without ECSI, we compare the secrecy rates of CR and CJ in high transmit power regimes. Building on this, we propose a novel hybrid scheme in which the relay utilizes both CR and CJ, and study the power allocation of the relay between CR and CJ for maximizing the secrecy rate under individual power constraints. Further, we study the case with imperfect channel state information (CSI) for both CR and CJ. At last, extensive numerical results are provided.

Mel Frequency Cepstral Coefficients (MFCC) are the most popular acoustic features used in automatic speech recognition (ASR), mainly because the coefficients capture the most useful information of the speech and fit well with the assumptions used in hidden Markov models. As is well known, MFCCs already employ several principles which have known counterparts in the peripheral properties of human hearing: decoupling across frequency, mel-warping of the frequency axis, log-compression of energy, etc. It is natural to introduce more mechanisms in the auditory periphery to improve the noise robustness of MFCC. In this paper, a k-nearest neighbors based frequency masking filter is proposed to reduce the audibility of spectra valleys which are sensitive to noise. Besides, Moore and Glasberg's critical band equivalent rectangular bandwidth (ERB) expression is utilized to determine the filter bandwidth. Furthermore, a new bandpass infinite impulse response (IIR) filter is proposed to imitate the temporal masking phenomenon of the human auditory system. These three auditory perceptual mechanisms are combined with the standard MFCC algorithm in order to investigate their effects on ASR performance, and a revised MFCC extraction scheme is presented. Recognition performances with the standard MFCC, RASTA perceptual linear prediction (RASTA-PLP) and the proposed feature extraction scheme are evaluated on a medium-vocabulary isolated-word recognition task and a more complex large vocabulary continuous speech recognition (LVCSR) task. Experimental results show that consistent robustness against background noise is achieved on these two tasks, and the proposed method outperforms both the standard MFCC and RASTA-PLP.

Conventional features for Automatic Speech Recognition and Sound Event Recognition such as Mel-Frequency Cepstral Coefficients (MFCCs) have been shown to perform poorly in noisy conditions. We introduce an auditory feature based on the gammatone filterbank, the Selective Gammatone Envelope Feature (SGEF), for Robust Sound Event Recognition where channel selection and the filterbank envelope is used to reduce the effect of noise for specific noise environments. In the experiments with Hidden Markov Model (HMM) recognizers, we shall show that our feature outperforms MFCCs significantly in four different noisy environments at various signal-to-noise ratios.

A new approach to speech feature estimation under noise circumstances is proposed in this paper. It is used in noise-robust continuous speech recognition (CSR). As the noise robust techniques in isolated word speech recognition, the running spectrum analysis (RSA), the running spectrum filtering (RSF) and the dynamic range adjustment (DRA) methods have been developed. Among them, only RSA has been applied to a CSR system. This paper proposes an extended DRA for a noise-robust CSR system. In the stage of speech recognition, a continuous speech waveform is automatically assigned to a block defined by a short time length. The extended DRA is applied to these estimated blocks. The average recognition rate of the proposed method has been improved under several different noise conditions. As a result, the recognition rates are improved up to 15% in various noises with 10 dB SNR.

Many researchers have proposed ultrasound imaging techniques for product inspection; however, most of these techniques are aimed at detecting the existence of flaws in products. The acquisition of an accurate three-dimensional image using ultrasound has the potential to be a useful product inspection tool. In this paper we apply the Envelope algorithm, which was originally proposed for accurate UWB (Ultra Wide-Band) radar imaging systems, to ultrasound imaging. We show that the Envelope algorithm results in image deterioration, because it is difficult for ultrasound measurements to achieve high signal to noise (S/N) ratio values as a result of a high level of noise and interference from the environment. To reduce errors, we propose two adaptive smoothing techniques that effectively stabilize the estimated image produced by the Envelope algorithm. An experimental study verifies that the proposed imaging algorithm has accurate 3-D imaging capability with a mean error of 6.1 µm, where the transmit center frequency is 2.0 MHz and the S/N ratio is 23 dB. These results demonstrate the robustness of the proposed imaging algorithm compared with a conventional Envelope algorithm.

In multiuser cognitive radio (CR) networks, we address the problem of joint transmit beamforming (BF) and power control (PC) for secondary users (SUs) when they are allowed to transmit simultaneously with primary users (PUs). The objective is to optimize the network sum rate under the interference constraints of PUs, which is a nonconvex problem. Iterative dual subgradient (IDuSuG) algorithm is proposed by iteratively performing BF and PC to optimize the sum rate, among which minimum mean square error (MMSE) or virtual power-weighed projection (VIP2) is used to design beamformers and subgradient method is used to control the power. VIP2 algorithm is devised for the case in which the interference caused by MMSE beamformer exceeds the threshold. Moreover, channel uncertainty due to lack of cooperation is considered. A closed-form worst-case expression is derived, with which the uncertainty optimization problem is transformed into a certain one. A robust algorithm based on IDuSuG is provided by modifying updates in iterative process. Furthermore, second-order cone programming approximation (SOCPA) method is proposed as another robust algorithm. Typical network models are approximated to SOCP problems and solved by interior-point method. Finally the network sum rates for different PU and SU numbers are assessed for both certainty and uncertainty channel models by simulation.

Telecommunications networks have become an important social infrastructure, and their robustness is considered to be a matter of social significance. Conventional network planning methods are generally based on the maximum volume of ordinary traffic and only assume explicitly specified failure scenarios. Therefore, present networks have marginal survivability against multiple failures induced by an extraordinarily high volume of traffic generated during times of natural disasters or popular social events. This paper proposes a telecommunications network planning method based on probabilistic risk assessment. In this method, risk criterion reflecting the degree of risk due to extraordinarily large traffic loads is predefined and estimated using probabilistic risk assessment. The probabilistic risk assessment can efficiently calculate the small but non-negligible probability that a series of multiple failures will occur in the considered network. Detailed procedures for the proposed planning method are explained using a district mobile network in terms of the extraordinarily large traffic volume resulting from earthquakes. As an application example of the proposed method, capacity dimensioning for the local session servers within the district mobile network is executed to reduce the risk criterion most effectively. Moreover, the optimum traffic-rerouting scheme that minimizes the estimated risk criterion is ascertained simultaneously. From the application example, the proposed planning method is verified to realize a telecommunications network with sufficient robustness against the extraordinarily high volume of traffic caused by the earthquakes.