Conference Abstracts

Group C: Computing and Transport Systems

Session 4: Computing and Transport Systems

4-1 Title Novel Circuit Techniques for High-Speed Intra-Chip Communication
Author(s) Takahiro Hanyu
Abstract This paper presents new circuit techniques for high-speed low-powerintra-chip communication. The use of multiple-valued datarepresentation and multiple-valued current-mode circuitry makes dataand control information superposed on dual-rail wires, and makes thehardware simple, which achieves asynchronous Gbps data transfer withless wire counts and less power dissipation. Moreover, the use ofnon-volatile storage devices makes storage functions compactlymerged into a logic-circuit plane, which can realize ultra-widememory band width. The efficiency of the proposed circuittechniques is discussed in some concrete examples with test-chipfabrication.

4-2 Title Analysis of Second-Order Modes of Linear Continuous-Time Systems under Positive-Real Transformations
Author(s) Masayuki Kawamata
Abstract This paper discusses the behavior of the second-order modes of inearcontinuous-time systems under variable transformations with positive-real functions. We first discuss the case of lossless positive-realtransformations, and show that the second-order modes are invariantunder any lossless positive-real transformation. We next considerthe case of general positive-real transformations, and reveal thatthe values of the second-order modes are decreased under any generalpositive-real transformation. We achieve the derivation of theseresults by describing the controllability/observability Gramians oftransformed systems, with the help of the lossless positive-reallemma, the positive-real lemma, and state-space formulation oftransformed systems.

4-3 Title New Frontiers in Optical Communication: Ultrahigh-speed transmission and coherent transmission
Author(s) Masataka Nakazawa
Abstract Two technologies are emerging from recent optical transmissionresearch. One is ultrahigh-speed OTDM transmission with advancedmodulation formats. The other is coherent optical transmission usingthe optical phase of a laser beam. As regards high-speedtransmission technology, I shall describe 160 Gbit/s distortion-freetransmission over 600 km using time-domain optical Fouriertransformation. As regards coherent transmission technology, I shalldescribe a QAM transmission using a frequency-stabilized fiber laseroperating at 1.55 ƒÊm. The laser was frequency-stabilized to oneof the absorption lines of C2H2 molecules. A20 Msymbol/s coherent QAM signal was successfully transmitted over525 km by using an optical PLL circuit.

4-4 Title Simultaneous Measurement System of Electromagnetic Field Using Parallel Modulated Probe Array
Author(s) Kunio Sawaya
Abstract A simultaneous measurement method at different locations usingparallel modulated probe array is proposed to measure theelectromagnetic field ragiated by antennas simultanteously. Eachmodulated probe is excited by a local signal with differentfrequency, so that the received IF signal contains differentfrequency components which indicate the relative magnitude of the RFsignal at positions of the probes, and all the IF components aremeasured simultaneously by a broad-band microwave receiver. A 3-Dmeasurement system of ragiation pattern based on the present methodis developed. The performance of the system is also demonstrated bymeasuring the radiation efficiency of antennas.

4-5 Title New Direction of Broadband Wireless Technology
Author(s) Fumiyuki Adachi and Eisuke Kudoh
Abstract Technical challenge for the realization of future wireless networksis two-fold: (a) to overcome the highly frequency-selective fadingchannel and (b) to significantly reduce the transmit power.Frequency-domain equalization (FDE) can take advantage of channelfrequency-selectivity and improve the transmission performance.Using FDE, either single carrier (SC) or multicarrier (MC) signaltransmission can be used for the downlink (base-to-mobile)applications. However, for the uplink (mobile-to-base) applications,SC transmission is more appropriate since it has less peak power.For broadband data transmissions, transmit power reduction is a veryimportant issue. Applying wireless multi-hop technique is a possiblesolution to this issue.

4-6 Title Long-Term CPU Load Prediction Method@for Scheduling of Distributed Parallel Processes
Author(s) Hirotomo Aso, Hiroshi Tatsumi and Yoshihiro Sugaya
Abstract There exist distributed processing environments composed of manyheterogeneous computers and it is required to schedule the distributedparallel processes in an appropriate manner. For the scheduling,prediction of execution load of a process is effective to exploit theresources of environments. We propose a long-term load predictionmethod with references of properties of processes and run timepredictions.

Author(s) Shozo Makino, Akinori Ito, Motoyuki Suzuki and Yuichi Ohkawa
Abstract Spoken language processing is very important for global-network-oriented electronics. In order to develop a high-accuracy spokenlanguage understanding system, in this COE-program we have carriedout the researches concerning to the following five topics. 1) Humanand machine communication system using speech, language and image:We have constructed various kinds of multi-modal communicationsystem such as a portable spoken dialog system, a gesture and facerecognition system, and a discrimination system among laugh, smileand normal, 2) English pronunciation and prosody evaluation systemfor Japanese speakers based on speech recognition, 3) Musicinformation retrieval system with query-by-song using error tolerantmelody matching: As we have not used lyrics for music informationretrieval yet, we introduced recognition of lyrics using speechrecognition., 4)Robust speech recognition system for various kindsof noisy speech: We have constructed a noise robust speechrecognition system in which each distribution of each node inphoneme HMM represents a single kind of noise, and 5)Topic-basedlanguage modeling using probabilistic latent semantic analysis(PLSA): In order to construct a fine language model, a languagemodel should be constructed for each topic, therefore we use PLSAfor clustering topics.

Go to TOP of this page