NAIST Graduate School of Information Science
| Date: | Thursday, Sep. 29, 2005 |
|---|---|
| Time: | 13:30 - 16:00 |
| Place: | L1 Lecture Room |
| Language: | English (Oral Presentation), English/Japanese (Question) |
| Chairperson: | Ryuhei TENMOKU and Sei IKEDA(Vision and Media Computing Lab. : D3) |
In this presentation, we propose a method of accelerating test generation for sequential circuits by using the knowledge about the availability of state justification sequences, the bound on the length of the distinguishing sequence, differentiation between valid and invalid states, and the existence of a reset state. The sequential circuit is synthesized from a given FSM description by the synthesis for testability (SfT) method which takes into consideration the features of our test generation method. The SfT method guarantees that the test generator will be able to find a state distinguishing sequence by modifying the FSM to a reduced FSM. The proposed method extracts the state justification sequence from the completely specified state transition function of the FSM produced by the synthesizer to improve the performance of its test generation process. Experimental results show that the proposed method can achieve 100% fault efficiency and identify every fault to be detectable or redundant in relatively short test generation time.
In this presention, we proposed a monopole-array assisted Doppler spread compensator with dummy elements for digital terrestrial television broadcasting receiver. Dummy elements placed at the either end of array antenna mitigate the performance degradation due to the mutual coupling amongst monopole-array antenna elements. The computer simulation results shows that the dummy elements is not only capable of reducing degradation due to mutual coupling, but also, it expands the operating bandwidth of the Doppler spread compensator with conventional mutual coupling canceller, which cancels out the mutual coupling in base band domain.
Orthogonal frequency division multiplexing (OFDM) is widely believed as an excellent technology that has a bright future in many next generations of wireless communications system e.g. 4G system. OFDM is robust against multipath fading effects as a consequence of longer symbol period and guard interval (GI) insertions. However, OFDM poses high peak-to-average power ratio (PAPR) problem which is a typical problem to all multicarrier systems. In this presentation, we propose a new Fourier(s) spreading as spreading technique to reduce the PAPR of OFDM while providing the high performance and high throughput. The original idea was proposed in 2005 [K. Anwar et. al. and Japanese Patent No. 2005-225604], while in this presentation we propose the improvement of the OFDM throughput double (2x) for obtaining higher bit rate. By the proposed technique, 2N data stream is possible to be transmitted on only N subcarriers without the requirement of bandwidth capacity increasing. For coveri ng this objective, we apply two Fouriers spreading at the transmitter, separator block coefficients for keeping the orthogonality between the data sets and two Fouriers despreading at the receiver. The results confirm that the proposed technique has gain by about 12dB bit-error-rate (BER) improvement in frequency selective fading channel and PAPR level reduced to 8dB from 12dB while the throughput is twice higher than the traditional OFDM system.
In this research, I will achieve reduction of a user's cost in the scientific simulation by Grid computing. In this presentation, I talk about following. 1. About the Grid system. 2. Trial and error in scientific simulation 3. About workflow system which supports performing scientific simulation. 4. Current research topic. It is necessary to do trial and error to perform the scientific simulation. The proposal workflow system automates processing necessary for the trial and error. As a result, it succeeded in decreasing the user's burden. Moreover, the research that improves the allocation of the computer resource is being done now. I explain the plan for the future about it.
With the rapid rise in the demand for location related service, communication devices such as PDAs or cellar phones must be able to search and manage information related to the geographical location. Leveraging locationrelated information is effective get an in-depth perspective on environmental circumstances, traffic situations and/or other problems. To handle the large number of information and queries communication devices create in the current ubiquitous environment, some scalable mechanism must be required. In this paper, we propose a peer-to-peer network system called “Mill” which can efficiently handle information related to the geographical location. To simplify the management of the location related information, we convert two dimensional coordinates into one dimensional circumference. Using this technique, Mill can search information by O(logN). Mill does not adopt any flooding method, and it reduces the amount of search queries compared with other P2P networks using flooding. DHT networks also do not leverage flooding and have good features. Simulation results show that the performance of Mill is good as well as other DHT networks. DHTs support only exact match lookups. The exact match is not suitable for searching information of a particular region. Mill provides an effective region search, by which users can search flexibly locationrelated information from small regions to large regions.
The aim of this ongoing study is to develop a remote system for quantitative analysis of PET images. PET scans using FDG have a potential to visualize detailed glucose metabolism in the brain. However, it is not practical because of the following reasons: one is a cumbersome arterial blood sampling, and the other is related with the difficulties in a voxel-by-voxel model estimation. Hence, less invasive and more rapid quantitative analysis is expected to be developed. The proposed method consists of two developing techniques: one is a method for estimating radioactive time curve in blood, and the other is voxel-by-voxel model estimation with a combination of MAP estimation and a clustering approach. As a result, the proposed algorithm formed physiologically acceptable images of glucose metabolism within a practical calculation time.
Security cameras are increasingly used for crime prevention. In most existing systems, images from the cameras are simply monitored by observers or recorded, and it takes a lot of time and costs to look over their images. In this presentation, I propose a networked security camera system for outdoor parking spaces, and demonstrate our prototype system for event detection. The prototype of event detector only uses fundamental image processing technique, but it can automatically detect cars and people coming and going without omission. And we can check their appearance from remote place using our prototype of event viewer.