Computer Logical Design Laboratory

Tsuneo Tsukahara Professor	Robert H. Fujii Associate Professor

The following researches are progressing in Computer Logical Design Laboratory: Tsuneo Tsukahara: Software-Defined Radio Receivers Related to this topic, the following work was done in 2007. A Spectrum-Analyzing Method for Multi-band Wireless Receivers: Recently the demand for wireless systems such as sensor networks has been rapidly growing. However, radio-wave resources are limited and invaluable especially in these days. Therefore, software-defied radios (SDRs) and cognitive radios, which is a principal application of SDR, can be the key to greatly improving frequency-spectrum efficiency. SDRs demand flexibility and reconfigurability in RF (Radio Frequency) circuits. Therefore, a frequency-efficient wireless transceiver architecture is indispensable. In 2006, we proposed a multi-band wireless receiver using a spectrum-analyzing method. In a receive mode, radio frequency (RF) signals are first amplified and converted to intermediate frequency (IF) signals less than about 10 MHz using mixers or frequency converters. In a spectrum-analyzing method, we use Fourier-series expansion of IF signals in the analog-signal domain because undesired strong signals should be attenuated before analog to digital (A/D) converters to save the number of resolution bits of A/D converters. Coefficients of the Fourier series are calculated using an analog multiplier and an integrator. Because IF frequencies of undesired strong signals are known, their coefficients are forced to zero, thereby being attenuated. Because Fourier-series coefficients of desired signals coming from different radio bands are simultaneously obtained, we can select suitable radio bands for stable communication. Robert Fujii: Neural Networks Due to the amazingly complex and numerous neural connections and the huge number of neurons present in the brain, only phenomenological system level notions about the workings of an entire mammalian brain are currently known. At the brain sub-system level, biologists and computational neural scientists have been somewhat successful at modeling brain regions such as the hippocampus and the auditory nerve system using mathematical models and simplified neuron models such as chaotic attractor and integrate and fire neuron?based neural networks. At the lowest brain system level, biologically and physically realistic brain modelers have used very detailed neuron models (e.g. Hodgkin and Huxley model) to simulate the behavior of small neural networks; however, the very fact that detailed physiologically realistic neuron models were used has made it difficult to build larger neural networks whose behavior could be analyzed properly. Research was carried out in three directions: Invariant learning using the hierarchical, parallel, and distributed organization model of the biological neocortex: a) the invariant recognition of object movement trajectories was investigated; b) the invariant recognition of multiple non-overlapped objects. Chaotic neural network sub-system for image recognition. Detailed biological neuron characteristics: a)the analysis of the mechanisms that allow rhythmic behavior such as in walking using coupled biological neuron oscillators; and b) the analysis of the action potential initiation site within a biological dendrite-neuron-axon structure as well as how the current from an action potential back-propagates into the dendrite. Future work in the neural networks area will include:a) development of a neural system capable of prediction; b) a mathematical model of the rhythmic coupling that occurs in biological neurons; and c)further studies in the structure of the neocortex. Robotics Research in the area of autonomous,intelligent,and interactive robotics for the home or in a work environment is gaining momentum. Research and development work in the areas of bipedal robot real-time control and image processing for object recognition were carried out. The bi-pedal robot that was used is approximately 50cm tall and is equipped with 22 servo motors, a camera, and a 3-axis accelerometer. A pico-size motherboard equipped with a 1 GHz Intel type processor, the Linux operating system with real-time patches, and a notebook PC connected serially to the pico-size motherboard were used for the control of the robot. Close to real-time image recognition was achieved by using two separate processors for image processing and for robot motion control. The notebook PC was used for image processing (OpenCV library) and the pico-size motherboard for robot servomotor control. The image processing data was sent to the motherboard using a serial connection. Bipedal robot locomotion ( fast and slow walk, turn, walking backwards, walking sideways) could be achieved with relative stability using multi-thread processing routines. Balancing in-place on a board that could be tilted in various directions was also achieved. Future work in the robotics area will include: a)the development of a fully autonomous bipedal robot that carries on its frame small-sized computers, sensors, and a battery; b) developing algorithms for soccer playing in order to compete in the RoboCup Soccer Tournament; and c) neural network based learning and recognition to be carried out on the small computers carried on the robot's frame. 164

Refereed Proceeding Papers

Books

[fujii-01:2007]	T. Ichishita and R.H. Fujii. Performance Evaluation of a Temporal Sequence Learning Spiking Neural Network. In Seventh International Convergence on Computer and Information Technology " ,, pages 616{620, Aizu Wakamatsu City, Fukushima, Japan, October 2007. CIT, IEEE Computer Society.
	The performance evaluation of a temporal sequence learning spiking neural network was carried out. Neural network characteristics that were evaluated included: long temporal sequence length recognition, factors that affect size of the neural network, and network robustness against random input noise. Music melodies of various lengths were used as temporal sequential input data for the evaluation. Results have shown that the spiking neural network can be made to learn inter-spike time sequences comprised of as many as 900 inter-spike times. The size of the neural network was in uenced by the amount and type of random noise used during the supervised learning phase. The spiking neural network system performance was approximately 90even in the presence of various types of random noise.
[fujii-02:2007]	R.H. Fujii and T. Ichishita. Effect of Synaptic Weight Assignment on Spiking Neuron Response. In The Sixth International Conference on Machine Learning and Applications " ,, pages 217{222, Cincinnati, Ohio, USA, December 2007. ICMLA, IEEE Computer Society.
	Three synaptic weight assignment schemes were proposed and their effect on the behavior of a spiking neuron were analyzed. Practical hardware implementation issues were also discussed.
[tsuka-01:2007]	T. Tsukahara. Invited Paper: RF CMOS Circuits Overview and Perspective. In 2007 Int `l Conference on Solid State Devices and Materials (SSDM), pages 752-753. JJAP, Sept. 2007,.
	The current trends in CMOS RF transceiver architectures are first discussed. Subsequently, a low-voltage-operation low-IF receiver using CMOS/SOI is described. Finally, the potentials of sampling techniques for RF signal processing are reviewed.

[tsuka-02:2007]	T. Tsukahara. Nikkei Electronics Tutorial: Introduction to CMOS RF Circuit Design,. Nikkei BP,, Dec., Jan., Feb., March, 2007-2008,.
	We describe basics on CMOS RF circuit design. The first volume presents background. Then, the second and third volumes explain frequency conversion and digital modulation schemes. Finally, the fourth volume describes some transceiver architectures.

Grants

[tsuka-03:2007]	T. Tsukahara. Commisioned Research Fund from Seiko Epson Corporation, 2007.
[tsuka-04:2007]	T. Tsukahara. Cooperative Research Fund from OKI Electric Industry Co., Ltd., 2007.
[tsuka-05:2007]	T. Tsukahara. Grants-in Aid for Scientic Research (KAKENHI) from JSPS [Young Researcher (Start-up)], 2006-2007.
[tsuka-06:2007]	T. Tsukahara. Commisioned Research Fund from ADVANTEST Corporation, 2007,.

Academic Activities

[tsuka-07:2007]	T. Tsukahara, 2007. Member of the Program Comittee, 2007 IEEE Asian Solid-State Circuits Conference (A-SSCC2007)
[tsuka-08:2007]	T. Tsukahara, 2007. Member of the Program Comittee, 2007 IEEE Int'l Conference on Computer and Information Technology (CIT2007)

Ph.D, Master and Graduation Theses

[fujii-03:2007]	Mayumi Abe. Graduation Thesis: Real-Time Interface for Robotics, University of Aizu, 2007. Thesis Advisor: Fujii, R.H.
[fujii-04:2007]	Tomoki Takahashi. Graduation Thesis: Vision-Based Self-Localization and Heading Angle Determination for Bipedal Robot Control, University of Aizu, 2007. Thesis Advisor: Fujii, R.H.
[fujii-05:2007]	Wakasa Edo. Graduation Thesis: Robot Walking Motion Implemented with Inverse Kinematics, University of Aizu, 2007. Thesis Advisor: Fujii, R.H.
[fujii-06:2007]	TKohei Fujiwara. Graduation Thesis: Character Recognition of Kana (Japanese Syllabary) by NuPIC, University of Aizu, 2007. Thesis Advisor: Fujii, R.H.
[fujii-07:2007]	Shoji Miura. Graduation Thesis: Neural Network Based on the Hodgkin- Huxley Biological Neuron Model, University of Aizu, 2007. Thesis Advisor: Fujii, R.H.
[fujii-08:2007]	Masaya Suda. Master Thesis: ZMP Based Adjustment for Bipedal Robot Walk, University of Aizu, 2007. Thesis Advisor: Fujii, R.H.
[tsuka-09:2007]	Hidekazu Usui. Graduation Thesis: A Study of a Frequency Synthesizer using a WIF Suitable for a Spectrum Analyzing Receiver, University of Aizu, 2007. Thesis Advisor: Tsukahara, T.
[tsuka-10:2007]	Daisuke Sato. Graduation Thesis: An Architectural Study on Wide-band and High Precision Receivers using a Windowed Integration Sampler (WIS), University of Aizu, 2007. Thesis Advisor: Tsukahara, T.
[tsuka-11:2007]	Akio Ogino. Graduation Thesis: Compensation Circuits for Broadband Windowed Integration Sampler (WIS) Receivers, University of Aizu, 2007. Thesis Advisor: Tsukahara, T.
[tsuka-12:2007]	Kengo Matsuyama. Graduation Thesis: A Design for Low-Noise and Low-Distortion Sampling Circuits, University of Aizu, 2007. Thesis Advisor: Tsukahara, T.
[tsuka-13:2007]	Tadahiro Yoshida. Graduation Thesis: A Signal Selecting Method Using Sampling and Fourier Transform, University of Aizu, 2007. Thesis Advisor: Tsukahara, T.

2009 THE UNIVERSITY OF AIZU ALL RIGHTS RESERVED