Empirical mode decomposition (EMD) is a newly developed tool to analyze nonlinear and non-stationary signals. It is used to decompose any signal into a finite number of time varying subband signals termed as intrinsic mode functions (IMFs). Such data adaptive decomposition is recently used in speech enhancement. This study presents the concept of EMD and its application to advanced speech signal processing paradigms including speech enhancement by soft-thresholding, voiced/unvoiced (V/Uv) speech discrimination and pitch estimation. The speech processing is frequently performed in the transformed domain and the transformation is usually achieved by traditional signal analysis techniques i.e. Fourier and wavelet transformations. These analysis methods employ priori basis function and it is not suitable for data adaptive analysis for non-stationary signal like speech. Recently, EMD is taken much attention for speech signal processing in data adaptive way. Several EMD based potential soft-thresholding algorithms for speech enhancement are discussed here. The V/Uv discrimination is an important concern in speech processing. It is usually performed by using acoustic features. The training data is used to determine the threshold for classification. The EMD based data adaptive thresholding approach is developed for V/Uv discrimination without any training phase. Noticeable improvement is achieved with the application of EMD in pitch estimation of noisy speech signals. The related experimental results are also presented to realize the effectiveness of EMD in advanced speech processing algorithms.

抄録全体を表示

Stochastic resonance (SR) is a phenomenon in which dynamic noise is effectively used to induce state transitions in a double-well potential system driven by subthreshold input signals. The noises are supplied to the system as an additional force. Recently, a phenomenon called "chaotic resonance" (CR) has been spotlighted in the literature. CR can be observed in chaotic systems that have multiple strange attractors and the ability to accept subthreshold input signals; i.e., such CR systems do not require any external noise source, unlike traditional SR systems. In this study, we employed Chua's oscillator as a candidate CR system. The oscillator was driven by a sinusoidal voltage source providing subthreshold input signals. In a certain range of input signal frequencies, we observed chaotic state transitions between the two attractors, whereas no state transition between the attractors was observed in the remaining frequency range. These findings indicated that chaotic fluctuations assisted the state transition. Furthermore, we observed nonmonotonic CR characteristics (correlation value and signal-to-noise ratio between the input signal and the output signal) that corresponded to typical nonmonotonic SR curves.

抄録全体を表示

The behavioral model of class E amplifiers in the steady state are given via the modified nodal analysis (MNA) formulation. The MNA formulation that provides circuit equations is adopted in general-purpose circuit simulators such as SPICE. Even though the circuit configuration is changed，the circuit equations are easily obtained. Hence, the behavioral model can be certainly obtained without difficulty in the setting up of the circuit equations. To produce the behavioral model, the MOSFET included in the class E amplifier is replaced with an ideal switch. Each dynamical system that depends on the "on" and "off" states of the switch is categorized as a descriptor system in control theory. The behavioral model of the class E amplifiers is obtained, on the basis of control theory. The proposed method is applicable to behavioral modeling of other types of resonant power amplifiers.

抄録全体を表示

Coordinate rotation digital computer (CORDIC) is an efficient algorithm for calculating various complex mathematical functions and has been widely used to accelerate embedded processors. The conventional CORDIC hardware cannot achieve high flexibility and high calculation precision simultaneously. This paper presents a programmable architecture of the CORDIC algorithm to calculate various complex mathematical functions with high flexibility, high calculation precision, and reasonable processing speed. The proposed programmable architecture of CORDIC employs an interpreter and a programmable CORDIC iterator. These approaches can dynamically self-regulate the hardware function for calculating 15 different mathematical functions without any hardware modification. The implemented results prove that the proposed architecture is suitable for accelerating an embedded processor.

抄録全体を表示

Blind source separation (BSS) is a technique for recovering original source signals from mixed signals without the aid on information of the source signals. The system restores the original source signals using the probability of the distribution of the original signal. In this paper, we consider the case where the original source signals are nonlinearly mixed. In general, the separation of the nonlinear mixed signals is difficult. In order to solve this problem, we apply a radial basis function (RBF) network with the nonlinear BSS system. The RBF network can approximate the nonlinear mapping. Therefore, the inverse mapping of the nonlinear mixture system is approximated by the RBF network. For the system to approximate the inverse mapping, it is necessary to adjust the parameters of the RBF network. We assume the original source signals to be independent of each other. In this case, if the mixed signals can be separated, the higher-order cross-moment of the output signals is decreased. In order to adjust the parameters of the RBF network, particle swarm optimization is used. We confirm the separation performance by numerical simulations. Simulation results indicate that the proposed approach has good performance.

抄録全体を表示

The optical zero-correlation zone (ZCZ) code, which is a set of pairs of binary and biphase sequences with a zero-correlation zone, can provide code division multiple access (CDMA) communication systems without cochannel interference. The M-ary amplitude shift keying spread spectrum (M-ary/ASK-SS) system using this code can detect a desired sequence without interference from undesired sequences. We have proposed the construction of a compact matched filter bank for this code, which is called the compact-type matched filter bank. However, the M-ary/ASK-SS system using this filter bank might have reduced bit error rate (BER) performance because the input-output characteristics of this filter bank for this code with the zero-correlation zone 4n - 2 with positive n are nonlinear. In this paper, we clarify that the BER performance of an M-ary/ASK-SS system using a compact-type matched filter bank over the additive white Gaussian noise (AWGN) and Rayleigh fading channels is not influenced by the nonlinear characteristics of this filter bank, using the theoretical formula of BER characteristics and computer simulation. The BER performance of the ASK-SS system using the compact-type matched filter is reduced compared with that of a system using a direct-type matched filter, as expected. However, the BER performance of the M-ary/ASK-SS system using this compact-type matched filter bank is equal to that of a system using a parallel-type matched filter bank. In fact, this compact-type matched filter bank is effective in the M-ary/ASK-SS system because the BER performance of the M-ary/ASK-SS system is not influenced by the nonlinear characteristics of the compact-type matched filter bank.

抄録全体を表示

In this paper, we investigate how to determine the subframe length and data transfer rate for packet length control with rate adaptation in wireless LANs. Specifically, in order to reduce packet delay (i.e., to shorten the time of packet transmission), we propose a novel cross-layer design for packet length and rate control, which are jointly optimized among all layers. In our framework, a packet is divided into several subframes, and their sizes are adaptively changed. Unlike the conventional decision scheme, our criterion is calculated from both users' desired packet delay and channel state information (CSI). Moreover, we consider the combination between packet length control and rate adaptation. To demonstrate the effectiveness of our framework, we evaluate what in terms of fragmentation overheads and packet delay by computer simulation and numerical calculation. As a result, our method can reduce the overhead cost of packet fragmentation by 34.0% and 79.4% and reduce packet delay by 18.4% and 0.714% for bit error probabilities pb of 10^-3 and 10^-5, respectively.

抄録全体を表示

It is well known that linear predictive coding (LPC) is sensitive to additive noise and that the corresponding noise compensation method suffers from poor accuracy of noise power estimation. In this paper, a novel noise compensation LPC method based on pitch synchronous analysis is presented. Using the proposed method, a more accurate estimation of noise power is calculated on each current frame. Experimental results show that the proposed method yields better performance than some conventional LPC analysis methods in the presence of white noise.

抄録全体を表示

This paper is concerned with the discrete combinatorial optimization problem. A modified hybrid algorithm that combines the max-min ant system and set-based particle swarm optimization (MSPSO) is proposed to solve the traveling salesman problem (TSP). To avoid a local solution, the cooperative mechanism among individuals is applied in MSPSO. MSPSO is based on set-based particle swarm optimization (SPSO). Before the execution of SPSO, max-min ant system (MMAS) is executed to accelerate the convergence of the algorithm. The results of the numerical experiments show that the proposed approach is effective for solving the TSP.

抄録全体を表示

Noise can significantly impact the effectiveness of digital image processing. In this paper, an improved structure-based Gaussian noise variance estimation algorithm is presented. This method first separates the image into blocks and calculates homogeneity measures of every block through the proposed masks, taking the image structure into account. Then, the most homogeneous blocks are selected using a new threshold. Finally, pixel value variances of all selected blocks are averaged to estimate the global noise variance for one image. Comparative experiments with a variety of images using the proposed method and original structure-oriented method are described, and the experimental results show that the proposed method is feasible and effective, especially for good-quality images.

抄録全体を表示