Xiaorong JING Zhengzhong ZHOU Tianqi ZHANG
We propose a feasible V-BLAST detector based on modified Householder QRD (M-H-QRD) over spatially correlated fading channel, which can almost match the performance of the V-BLAST algorithm with much lower complexity and better numerical stability. Compared to the sorted QRD (S-QRD) detector, the proposed detector requires a smaller minimum word-length to reach the same value of error floor for fixed-point (FP) numerical precision despite no significant performance difference for floating-point machine precision. All these advantages make it attractive when implemented using FP arithmetic.
Ravi GONDHALEKAR Jun-ichi IMURA
A method for the exact cost performance analysis of autonomous discrete-time piecewise affine systems is presented. Cost performance refers to the average trajectory cost over the entire region of attraction of the origin. The trajectory cost is based on the infinite-horizon cost. First, a reverse reachability algorithm which constructs the explicit piecewise quadratic trajectory cost function over the entire region of attraction of the origin is presented. Then, an explicit expression for the integral of a quadratic function over a simplex of arbitrary dimension is derived. Thus, the piecewise quadratic trajectory cost function can be integrated exactly and efficiently in order to determine the cost performance of the system as a whole. This alleviates the need to perform a large number of simulations.
This paper investigates the problem of nonpreemptively scheduling independent parallel tasks in an environment with multiple machines, which is motivated from the recent studies in scheduling tasks in a multi-machine environment. In this scheduling environment, each machine contains a number of identical processors and each parallel task can simultaneously require a number of processors for its processing in any single machine. Whenever tasks are processed in parallel in a parallel machine, message communication among processors is often inevitable. The problem of finding a shortest schedule length on scheduling independent parallel tasks with the consideration of communication overhead in a multi-machine environment is NP-hard. The aim of this paper is to propose a heuristic algorithm for this kind of problem and to analyze the performance bound of this heuristic algorithm.
Ryo NAGATA Jun-ichi KAKEGAWA Hiromi SUGIMOTO Yukiko YABUTA
This paper describes a method for recognizing romanized Japanese words in learner English. They become noise and problematic in a variety of systems and tools for language learning and teaching including text analysis, spell checking, and grammatical error detection because they are Japanese words and thus mostly unknown to such systems and tools. A problem one encounters when recognizing romanized Japanese words in learner English is that the spelling rules of romanized Japanese words are often violated. To address this problem, the described method uses a clustering algorithm reinforced by a small set of rules. Experiments show that it achieves an F-measure of 0.879 and outperforms other methods. They also show that it only requires the target text and an English word list of reasonable size.
The development of network technology reveals the clear trend that mobile devices will soon be equipped with more and more network-based functions and services. This increase also results in more intrusions and attacks on mobile devices; therefore, mobile security mechanisms are becoming indispensable. In this paper, we propose a novel signature matching scheme for mobile security. This scheme not only emphasizes a small resource requirement and an optimal scan speed, which are both important for resource-limited mobile devices, but also focuses on practical features such as stable performance, fast signature set updates and hardware implementation. This scheme is based on the finite state machine (FSM) approach widely used for string matching. An SRAM-based two-level finite state machine (TFSM) solution is introduced to utilize the unbalanced transition distribution in the original FSM to decrease the memory requirement, and to shorten the critical path of the single-FSM solution. By adjusting the boundary of the two FSMs, optimum memory usage and throughput are obtainable. The hardware circuit of our scheme is designed and evaluated by both FPGA and ASIC technology. The result of FPGA evaluation shows that 2,168 unique patterns with a total of 32,776 characters are stored in 177.75 KB SelectRAM blocks of Xilinx XC4VLX40 FPGA and a 3.0 Gbps throughput is achieved. The result of ASIC evaluation with 180 nm-CMOS library shows a throughput of over 4.5 Gbps with 132 KB of SRAM. Because of the small amount of memory and logic cell requirements, as well as the scalability of our scheme, higher performance is achieved by instantiating several signature matching engines when more resources are provided.
Autonomous Decentralized System (ADS) has been making progress in these 31 years since it was proposed in 1977. During these long years in the rapidly advancing computer and communication technologies, the ADS concept has not been changed but its technologies have been growing in accordance with the change and diversity of the social, economical and personal requirements and through the globalization of the market and the restructuring organizations. The ADS technologies are systematized to cover all processes of system design, operation, maintenance and modification. This paper reviews the work done in fields of ADS in past 31 years from not only technological perspectives, but it also encompasses users requirements and value, system design, industrial activity, academic activity and standardization [1]-[26]. Moreover the new directions of the ADS are suggested.
Tae-young KIM Jun PARK Cheol-Su LIM
In this paper, a motion evaluation technique for rehabilitation training is introduced. Motion recognition technologies have been developed for determining matching motions in the training set. However, we need to measure how well and how much of the motion has been followed for training motion evaluation. We employed a Finite State Machine as a framework of motion evaluation. For similarity analysis, we used weighted angular value differences although any template matching algorithm may be used. For robustness under illumination changes, IR LED's and cameras with IR-pass filter were used. Developed technique was successfully used for rehabilitation training of the disabled. Therapists appraised the system as practically useful.
Jaehoon KIM Youngsoo KIM Seog PARK
Recently, for more efficient filtering of XML data, YFilter system has been suggested to exploit the prefix commonalities that exist among path expressions. Sharing the prefix commonality gives the benefit of improving filtering performance through the tremendous reduction in filtering machine size. However, exploiting the postfix commonality can also be useful for an XML filtering situation. For example, when a stream of XML messages does not have any defined schema, or users cannot remember the defined schema exactly, users often use the partial matching path queries which begins with the descendant axis ("//"), e.g., '//science/article/title', '//entertainment/article/title', and '//title'. If so, the registered XPath queries are most likely to have the postfix commonality, e.g., the sample queries share the partial path expressions 'article/title' and 'title'. Therefore, in this paper, we introduce a bottom-up filtering approach exploiting the postfix commonality against the top-down approach of YFilter exploiting the prefix commonality. Some experimental results show that our method has better filtering performance when registered XPath queries mainly consist of the partial matching path queries with the postfix commonality.
Yoshimichi FUKASAWA Kiyoshi KAWAGUCHI Takashi YOSHIDA Takahiro SUGIYAMA Atsushi NAKAGAWA
A 76-GHz Gunn voltage-controlled oscillator (VCO) with a high output power and a wide tuning-frequency range was fabricated by optimizing VCO circuits and using laser micromachining. The tuning-frequency range of the fabricated Gunn VCO was more than two times higher than that attained in our previous experiments by optimizing VCO circuits. The VCO attained a tuning-frequency range of 493 MHz, output power variation of 1.0 dB, and tuning-frequency linearity of 6.1% over a tuning-voltage range from 0 to 10 V. Its power consumption was 2.0 W at operation voltage of 3.6 V. And it measured output power was 13.3 dBm with DC-RF conversion efficiency of 1.0% at 76.5 GHz. Moreover, under fundamental-mode operation, it achieved low phase noise of -107.8 dBc/Hz at an offset frequency of 1 MHz. Since laser micromachining was used in fabricating the Gunn VCO, the reproducibility of its RF performance was improved.
In this paper, we propose a midpoint-validation method which improves the generalization of Support Vector Machine. The proposed method creates midpoint data, as well as a turning adjustment parameter of Support Vector Machine using midpoint data and previous training data. We compare its performance with the original Support Vector Machine, Multilayer Perceptron, Radial Basis Function Neural Network and also tested our proposed method on several benchmark problems. The results obtained from the simulation shows the effectiveness of the proposed method.
Soong Hyun SHIN Sung Woo CHUNG Eui-Young CHUNG Chu Shik JHON
As technology scales down, leakage energy accounts for a greater proportion of total energy. Applying the drowsy technique to a cache, is regarded as one of the most efficient techniques for reducing leakage energy. However, it increases the Soft Error Rate (SER), thus, many researchers doubt the reliability of the drowsy technique. In this paper, we show several reasons why the instruction cache can adopt the drowsy technique without reliability problems. First, an instruction cache always stores read-only data, leading to soft error recovery by re-fetching the instructions from lower level memory. Second, the effect of the re-fetching caused by soft errors on performance is negligible. Additionally, a considerable percentage of soft errors can occur without harming the performance. Lastly, unrecoverable soft errors can be controlled by the scrubbing method. The simulation results show that the drowsy instruction cache rarely increases the rate of unrecoverable errors and negligibly degrades the performance.
Humans have a remarkable ability to flexibly control various objects such as tools. Much evidence suggests that the internal models acquired in the central nervous system (CNS) support flexible control. Internal models are neural mechanisms that mimic the input-output properties of controlled objects. In a series of functional magnetic resonance imaging (fMRI) studies, we demonstrate how the CNS acquires and switches internal models for dexterous use of many tools. In the first study, we investigated human cerebellar activity when human subjects learned how to use a novel tool (a rotated computer mouse, where the cursor appears in a rotated position) and found that activity reflecting an internal model of the novel tool increases in the lateral cerebellum after learning how to use the tool. In the second study, we investigated the internal-model activity after sufficient training in the use of two types of novel tools (the rotated mouse and a velocity mouse, where the cursor's velocity is proportional to mouse's position) and found that the cerebellar activities for the two tools were spatially segregated. In the third study, we investigated brain activity associated with the flexible switching of tools. We found that the activity related to switching internal models was in the prefrontal lobe (area 46 and the insula), the parietal lobe, and the cerebellum. These results suggest that internal models in the cerebellum represent input-output properties of the tools as modulators of continuous signals. The cerebellar abilities in adaptive modulation of signals can be used to enhance the control signals in communications between the brain and computers.
Ryohei P. HASEGAWA Yukako T. HASEGAWA Mark A. SEGRAVES
To examine the function of the superior colliculus (SC) in decision-making processes and the application of its single trial activity for "neural mind reading," we recorded from SC deep layers while two monkeys performed oculomotor go/no-go tasks. We have recently focused on monitoring single trial activities in single SC neurons, and designed a virtual decision function (VDF) to provide a good estimation of single-dimensional decisions (go/no-go decisions for a cue presented at a specific visual field, a response field of each neuron). In this study, we used two VDFs for multidimensional decisions (go/no-go decisions at two cue locations) with the ensemble activity which was simultaneously recorded from a small group (4 to 6) of neurons at both sides of the SC. VDFs predicted cue locations as well as go/no-go decisions. These results suggest that monitoring of ensemble SC activity had sufficient capacity to predict multidimensional decisions on a trial-by-trial basis, which is an ideal candidate to serve for cognitive brain-machine interfaces (BMI) such as two-dimensional word spellers.
Asifullah KHAN Syed Fahad TAHIR Tae-Sun CHOI
We present a novel approach to developing Machine Learning (ML) based decoding models for extracting a watermark in the presence of attacks. Statistical characterization of the components of various frequency bands is exploited to allow blind extraction of the watermark. Experimental results show that the proposed ML based decoding scheme can adapt to suit the watermark application by learning the alterations in the feature space incurred by the attack employed.
Hideo OKUMA Hirofumi YAMAMOTO Eiichiro SUMITA
This paper presents a method to effectively introduce a translation dictionary into phrase-based SMT. Though SMT systems can be built with only a parallel corpus, translation dictionaries are more widely available and have many more entries than parallel corpora. A simple and low-cost method to introduce a translation dictionary is to attach a dictionary entry into a phrase table. This, however, does not work well. Target word order and even whole target sentences are often incorrect. To solve this problem, the proposed method uses high-frequency words in the training corpus. The high-frequency words may already be trained well; in other words, they may appear in the phrase table and therefore be translated with correct word order. Experimental results show the proposed method as far superior to simply attaching dictionary entries into phrase tables.
Suresh M. NISSANKA Ken MISHINA Akihiro MARUTA Shunsuke MITANI Kazuyuki ISHIDA Katsuhiro SHIMIZU Tatsuo HATTA Ken-ichi KITAYAMA
All-optical wavelength conversion and modulation format conversion will be needed in the next generation high-speed optical communication networks. We have proposed and successfully demonstrated the error free operation of all-optical modulation format conversion from NRZ-OOK to RZ-BPSK using SOA based MZI wavelength converter. In this paper, we experimentally investigate the wavelength conversion characteristics of the proposed NRZ-OOK/RZ-BPSK modulation format converter. The results show that error free modulation format conversion is possible over the entire C band.
Q-Haing JO Yun-Sik PARK Kye-Hwan LEE Joon-Hyuk CHANG
In this letter, we propose effective feature vectors to improve the performance of voice activity detection (VAD) employing a support vector machine (SVM), which is known to incorporate an optimized nonlinear decision over two different classes. To extract the effective feature vectors, we present a novel scheme that combines the a posteriori SNR, a priori SNR, and predicted SNR, widely adopted in conventional statistical model-based VAD.
Preeti SHARMA Shiban K. KOUL Sudhir CHANDRA
This letter reports a high-performance Ka-band equilateral triangular microstrip patch (ETMP) antenna suspended on a thin dielectric membrane. The membrane is released using a silicon bulk-micromachining technique. A set of closed-form expressions to calculate the resonant frequency of the proposed antenna on the micromachined substrate is also presented. The measured performance of the antenna structure is verified using the finite element method (FEM) based Agilent High Frequency Structure Simulator (version 5.5). The fabricated antenna exhibited a wide -10 dB return loss bandwidth of 1.2 GHz at 35.4 GHz. The measured antenna cross-polarization level is less than -15 dB in both the E- and H-planes.
Dukyun NAM Dongman LEE Han NAMGOONG
We propose an efficient reachability estimation scheme for group membership services in mobile ad hoc networks (MANETs). The periodical message exchange-based scheme, i.e., a typical reachability estimation scheme, requires message exchanges even when the reachability status does not change. It is presumed that the reachability between nodes is maintained while the nodes move around in a limited range. The proposed scheme exploits a virtual grid for the course-grained estimation. A region in the virtual grid can be used to represent the movement range which does not change the reachability. Each node calculates how long it will stay in a region, and issues the duration information only when it gets out of the current region.
Spoken language understanding (SLU) aims to map a user's speech into a semantic frame. Since most of the previous works use the semantic structures for SLU, we verify that the structure is useful even for noisy input. We apply a structured prediction method to SLU problem and compare it to an unstructured one. In addition, we present a combined method to embed long-distance dependency between entities in a cascaded manner. On air travel data, we show that our approach improves performance over baseline models.