Existing standard speech coders can provide high quality speech communication. However, they tend to degrade the performance of automatic speech recognition (ASR) systems that use the reconstructed speech. The main cause of the degradation is in that the linear predictive coefficients (LPCs), which are typical spectral envelope parameters in speech coding, are optimized to speech quality rather than to the performance of speech recognition. In this paper, we propose a speech coder using mel-frequency cepstral coefficients (MFCCs) instead of LPCs to improve the performance of a server-based speech recognition system in network environments. To develop the proposed speech coder with a low-bit rate, we first explore the interframe correlation of MFCCs, which results in the predictive quantization of MFCC. Second, a safety-net scheme is proposed to make the MFCC-based speech coder robust to channel errors. As a result, we propose an 8.7 kbps MFCC-based CELP coder. It is shown that the proposed speech coder has a comparable speech quality to 8 kbps G.729 and the ASR system using the proposed speech coder gives the relative word error rate reduction by 6.8% as compared to the ASR system using G.729 on a large vocabulary task (AURORA4).

We propose a gradient-limited affine projection algorithm (GL-APA), which can achieve fast and double-talk-robust convergence in acoustic echo cancellation. GL-APA is derived from the M-estimation-based nonlinear cost function extended for evaluating multiple error signals dealt with in the affine projection algorithm (APA). By considering the nonlinearity of the gradient, we carefully formulate an update equation consistent with multiple input-output relationships, which the conventional APA inherently satisfies to achieve fast convergence. We also newly introduce a scaling rule for the nonlinearity, so we can easily implement GL-APA by using a predetermined primary function as a basis of scaling with any projection order. This guarantees a linkage between GL-APA and the gradient-limited normalized least-mean-squares algorithm (GL-NLMS), which is a conventional algorithm that corresponds to the GL-APA of the first order. The performance of GL-APA is demonstrated with simulation results.

Development of ultrafast optical interfaces that can operate in sub-terahertz region is important to apply superconducting electronic devices to the high-end systems. We have performed several fundamental researches to realize the ultrafast optical input interface for superconducting electronic devices. Firstly, we observed optical response of amorphous Ge thin films, and the results indicated that an amorphous Ge photoconductive switch could stably operate in a terahertz frequency range as an optical-to-electrical signal converter in the low-temperature region below Tc of YBCO. Next, we have fabricated optical-to-electrical signal conversion system with photomixing technique, and we have demonstrated the generation and the detection of high frequency signals over 50 GHz. Finally, we have observed optical responses of a Josephson vortex flow transistor under irradiation of femtosecond laser pulses, and the results suggeste that the device has high potential as an optical interface.

This paper proposes high-Q distributed constant passive devices using wafer-level chip scale package (WL-CSP) technology, which can be realized on a Si CMOS chip. A 90directional coupler using the WL-CSP technology has center frequency of 25.6 GHz, insertion loss of -0.5 dB and isolation of -29.8 dB in the measurement result. The WL-CSP technology contributes to realize low-loss RF passive devices on Si CMOS chip, which is indispensable to achieve small-size, cost-effective and low-power monolithic wireless communication circuits (MWCCs).

In this letter, we study the blocking probabilities in an asynchronous optical packet/burst switching system with full wavelength conversion. Most of the existing work use Poisson traffic models that is well-suited for an infinite population of users. In this letter, the optical packet traffic arriving at the switching system is modeled through a superposition of a finite number of identical on-off sources. We propose a block tridiagonal LU factorization algorithm to efficiently solve the two dimensional Markov chain that arises in the modeling of the switching system.

Ultrahigh-speed fiber lasers operating at up to 40 GHz offer a clean longitudinal comb and a narrow linewidth. This makes them suitable for applications including optical comb generation, ultrahigh-speed optical pulse transmission including PSK, and as opto-microwave oscillators. In this paper, we describe recent progress on ultrafast fiber lasers and their applications to optical metrology.

Traffic sign detection is a valuable part of future driver support system. In this paper, we present a novel framework to accurately detect traffic signs from a single color image by analyzing geometrical, physical and text/symbol features of traffic signs. First, we utilize an elaborate edge detection algorithm to extract edge map and accurate edge pixel gradient information. Then, we extract 2-D geometric primitives (circles, ellipses, rectangles and triangles) efficiently from image edge map. Third, the candidate traffic sign regions are selected by analyzing the intrinsic color features, which are invariant to different illumination conditions, of each region circumvented by geometric primitives. Finally, a text and symbol detection algorithm is introduced to classify true traffic signs. Experimental results demonstrated the capabilities of our algorithm to detect traffic signs with respect to different size, shape, color and illumination conditions.

We consider the bandwidth optimization problem in a Generalized Processor Sharing (GPS) server to minimize the total bandwidth such that QoS requirements for each class queue are satisfied. Our previous optimization algorithm [6] requires rather long optimization time to solve the problem. We propose a new optimization algorithm based on weight vector adjustment. Numerical results show that the required time to find the optimal resource in GPS servers is significantly reduced, compared to the previous algorithm.

A micromirror for an external cavity diode laser is described. The mirror is supported by two sets of parallel torsion bars enabling piston motion as well as rotation. These motions are for realizing continuous wavelength tuning. Adjusting two rotations electrically, the pivot of the mirror rotation can be controlled. The long stroke of the vertical comb is realized by the deep three-dimensional structure prepared by the wafer bending method.

An electrostatically driven micromirror is described. The vertical comb of a three-dimensional microstructure is realized by bending the device wafer having microstructures. By resetting the bending angle, the tuning of the vertical gap between moving and stationary combs is possible. The characteristics of the vertical comb drive actuator can be tuned, confirming the performance.

In this study we have employed an effective technique for dosimetric analyses of base station antennas in an underground environment. The technique combines a ray-tracing method and the finite-difference time-domain (FDTD) method to calculate the specific absorption rate (SAR) in the human body. The ray-tracing method was applied to evaluate the incident fields in relation to the exposed subject in a three-dimensional space, while the FDTD method was used to calculate the detailed SAR distributions in the human body. A scenario under an underground passage with the installation of a top-loaded monopole antenna was analyzed to investigate the relationship between the actual antenna exposure and a plane-wave exposure. The results show that the plane-wave exposure overestimated the whole-body average SAR in most cases, although this was not always true for peak SAR. The finding implies not only the usefulness of the present uniform-exposure-based reference level for the whole-body average SAR evaluation but also the necessity of modeling actual underground environment for high-precision local peak SAR evaluation.

This paper presents the characteristics and modeling of VoIP traffic for a real network. The new model, based on measured data, shows a significant difference from the previously proposed models in terms of parameters and their effects. It is found that the effects of background noise and ringing tones have essential influences on the model. The observed distributions of talkspurt and silent durations have long-tail characteristics and considerably differ from the existing models. An additional state called "Long burst", which represents the background noise at the talker's place, is added into the continuous-time Markov process model. The other three states, "Talk", "Short silence" and "Long silence", represent the normal behavior of the VoIP user. Models for conversational speech containing the communication during the dialogue are presented. In the case of the VoIP traffic aggregation, the simplified models, which neglect the conversation's interaction, are proposed. Depending on the occurrences of background noise during the speech, the model is classified as "noisy speech" or "noiseless speech". The measured data shows that the background noise typically increases the data rate by 60%. Simulation results of aggregated VoIP traffic indicate the self-similarity, which is analogous to the measured data. Results from the measurements support the fact that except the ringing duration the conversations from both the directions can be modeled in identical manner.

In this paper, we propose a novel open-loop Automatic Frequency Control (AFC) circuit suitable for 64QAM point-to-multipoint (P-MP) burst communications. The proposed AFC contains two frequency offset detectors. One estimates the phase rotation over long intervals to obtain accurate estimates at the cost of phase ambiguity. The other estimates the phase rotation over short intervals and its output is used to resolve the ambiguity in the following phase ambiguity compensator. Thus, the proposed AFC circuit calculates the phase rotation over sufficiently long periods to yield accurately estimate the carrier frequency offset while suppressing the phase-unwrapping problem. The proposed AFC approaches the Cramer-Rao bound (CRB) and so achieves very small residual frequency offset. The proposed AFC circuit can be implemented with much smaller circuit scale than the conventional devices. Computer simulations and experiments confirm that its residual frequency error is less than of 10-5 for the frame format considered; this performance is sufficient for the 64QAM -40 Mbaud system targeted.

A well-defined test structure of organic static-induction transistor (SIT) having regularly sized nano-apertures in the gate electrode has been fabricated by colloidal lithography using 130-nm-diameter polystyrene spheres as shadow masks during vacuum deposition. Transistor characteristics of individual nano-apertures, namely 'nano-SIT,' have been measured using a conductive atomic-force-microscope (AFM) probe as a movable source electrode. Position of the source electrode is found to be more important to increase current on/off ratio than the distance between source and gate electrodes. Experimentally obtained maximum on/off ratio was 710 (at VDS = -4 V, VGS = 0 and 2 V) when a source electrode was fixed at the edge of gate aperture. The characteristics have been then analyzed using semiconductor device simulation by employing a strongly non-linear carrier mobility model in the CuPc layer. From device simulation, source current is found to be modulated not only by a saddle point potential in the gate aperture area but also by a pinch-off effect near the source electrode. According to the obtained results, a modified structure of organic SIT and an adequate acceptor concentration is proposed. On/off ratio of the modified organic SIT is expected to be 100 times larger than that of a conventional one.

In this paper, we present preliminary work on recognizing affect from a Korean textual document by using a manually built affect lexicon and adopting natural language processing tools. A manually built affect lexicon is constructed in order to be able to detect various emotional expressions, and its entries consist of emotion vectors. The natural language processing tools analyze an input document to enhance the accuracy of our affect recognizer. The performance of our affect recognizer is evaluated through automatic classification of song lyrics according to moods.

A novel cache-based network processor (NP) architecture that can catch up with next generation 100-Gbps packet-processing throughput by exploiting a nature of network traffic is proposed, and the prototype is evaluated with real network traffic traces. This architecture consists of several small processing units (PUs) and a bit-stream manipulation hardware called a burst-stream path (BSP) that has a special cache mechanism called a process-learning cache (PLC) and a cache-miss handler (CMH). The PLC memorizes a packet-processing method with all table-lookup results, and applies it to subsequent packets that have the same information in their header. To avoid packet-processing blocking, the CMH handles cache-miss packets while registration processing is performed at the PLC. The combination of the PLC and CMH enables most packets to skip the execution at the PUs, which dissipate huge power in conventional NPs. We evaluated an FPGA-based prototype with real core network traffic traces of a WIDE backbone router. From the experimental results, we observed a special case where the packet of minimum size appeared in large quantities, and the cache-based NP was able to achieve 100% throughput with only the 10%-throughput PUs due to the existence of very high temporal locality of network traffic. From the whole results, the cache-based NP would be able to achieve 100-Gbps throughput by using 10- to 40-Gbps throughput PUs. The power consumption of the cache-based NP, which consists of 40-Gbps throughput PUs, is estimated to be only 44.7% that of a conventional NP.

Emerging GMPLS (Generalized Multi-Protocol Label Switching)-based photonic networks are expected to realize the dynamic allocation of network resources for a wide range of applications, such as carriers' backbone networks as well as enterprise core networks and GRID computing. To address diverse reliability requirements corresponding to different application needs, photonic networks have to support various optical path recovery schemes. Thus GMPLS standardization bodies have developed failure recovery protocols for 1+1 protection, 1:N protection and restoration, with support of extra traffic and shared use of back-up resources. Whereas the standardization efforts cover a full spectrum of recovery schemes, there have not been many reports on actual implementations of such functionalities, and none of them included extra traffic. This paper introduces an OXC (Optical Cross Connect) implementation of GMPLS's failure recovery functionalities supporting 1+1 protection, M:N protection and extra path. Here extra path is an extension of GMPLS protection's extra traffic which can partially reuse protected paths' back-up resources. Evaluation of the implementation confirms rapid recovery of protected traffic upon a failure, even when preemption of an extra path is involved. It is also shown that its preemption scheme can resolve the issue of the poor scalability of GMPLS-based preemption when multiple extra paths are preempted upon a failure.

In this paper, we propose a method to estimate affine motion parameters from consecutive images with the assumption that the motion in progress can be characterized by an affine model. The motion may be caused either by a moving camera or moving object. The proposed method first extracts motion vectors from a sequence of images and then processes them by adaptive robust estimation to obtain affine parameters. Typically, a robust estimation filters out outliers (velocity vectors that do not fit into the model) by fitting velocity vectors to a predefined model. To filter out potential outliers, our adaptive robust estimation defines a flexible weight function based on a sigmoid function. During the estimation process, we tune the sigmoid function gradually to its hard-limit as the errors between the input data and the estimation model are decreased, so that we can effectively separate non-outliers from outliers with the help of the finally tuned hard-limit form of the weight function. The experimental results show that the suggested approach is very effective in estimating affine parameters.

The sites that doped divalent Eu ions occupy in BaMgAl10O17 were studied by X-ray absorption fine structure (XAFS) measurement. The radial structural function and the Fourier-filtered EXAFS wiggle derived from the observed XAFS spectrum suggested that Eu ions occupy the Beevers-Ross and/or anti-Beevers-Ross sites. Observed XANES spectrum also could be explained by Beevers-Ross site occupation.

Distributed denial-of-service attacks on public servers have recently become more serious. More are SYN Flood attacks, since the malicious attackers can easily exploit the TCP specification to generate traffic making public servers unavailable. To assure that network services will not be interrupted, we need faster and more accurate defense mechanisms against malicious traffic, especially SYN Floods. One of the problems in detecting SYN Flood traffic is that server nodes or firewalls cannot distinguish the SYN packets of normal TCP connections from those of SYN Flood attack. Moreover, since the rate of normal network traffic may vary, we cannot use an explicit threshold of SYN arrival rates to detect SYN Flood traffic. In this paper we introduce a mechanism for detecting SYN Flood traffic more accurately by taking into consideration the time variation of arrival traffic. We first investigate the statistics of the arrival rates of both normal TCP SYN packets and SYN Flood attack packets. We then describe our new detection mechanism based on the statistics of SYN arrival rates. Our analytical results show that the arrival rate of normal TCP SYN packets can be modeled by a normal distribution and that our proposed mechanism can detect SYN Flood traffic quickly and accurately regardless of time variance of the traffic.