High-speed low-power matched filter plays an important role in the fast despreading of spread-signals in wideband code division multiple access (W-CDMA) mobile communications. In this paper, we describe the algorithm and the VLSI-architecture of a complex matched filter chip implemented by our proposed digital-controlled analog parallel operational circuits. The complex matched filter VLSI with variable taps from 4 to 128 is developed for despreading QPSK-modulated spread-signals for W-CDMA communications, which is fabricated by a 2-metal 0.8 µm CMOS technology. The dissipation power of the chip is 225 mW and 130 mW when it operates at the chip-rate of 20 MHz with the supply voltages of 3.0 V and 2.5 V, respectively, and it can be furthermore reduced to 62 mW at chip rate of 10 MHz when the supply voltage is lowered to 2.2 V. The 3-dB cut-off frequency of the fabricated chip is higher than 20 MHz for both 3.0 V and 2.5 V supplies. Comparing to pure digital matched filters, the massive and high-speed despreading operations of the spread-signals are directly carried out in analog domain. As a result, two high-speed analog-to-digital (A/D) converters operating at chip rate are omitted, the inner signal paths and the total dissipation power are greatly reduced.

The noise level distribution owing to only a non-stationary working objective machine has been stochastically expressed by reflecting the temporal change of distribution parameters under a generalized regression model especially with aid of the vibration level observation. The proposed method has been applied to a noise evaluation of non-stationarily operated jigsaw.

We propose an untraceable electronic money system. Our system uses the partially blind signature based on the discrete logarithm problem, and applies secret key certificates to the payment protocol.

A hierarchical structure of the statistical models involving the parametric, state space, generalized state space, and self-organizing state space models is explained. It is shown that by considering higher level modeling, it is possible to develop models quite freely and then to extract essential information from data which has been difficult to obtain due to the use of restricted models. It is also shown that by rising the level of the model, the model selection procedure which has been realized with human expertise can be performed automatically and thus the automatic processing of huge time series data becomes realistic. In other words, the hierarchical statistical modeling facilitates both automatic processing of massive time series data and a new method for knowledge discovery.

With the advent of digital video and digital broadcasting, copyright protection of video data has been one of the most important issues. We present in this paper a novel method of digital watermark for video data based on the discrete wavelet transform. In this method, we embed the watermark in the lowest frequency components of each frame in the uncoded video by using a controlled quantization process. The watermark can be extracted directly from the decoded video without access to the original video. Experimental results show that the proposed method gives the watermarked image of better quality and is robust against MPEG coding and re-encoding. Furthermore, we discuss multiple watermarking with regard to the generational copy control for video contents.

The abilities of fuzzy inference methods in modeling of complicated systems are implemented to electromagnetics for the first time. The very popular and well known monopole antenna is chosen as a general example and a fast, simple and accurate fuzzy model for its input impedance is made by introducing a new point of view to impedance basic parameters. It is established that a surprisingly little number of input data points is sufficient to make a full model and also the system behavior (dominant rules) are saved as simple membership functions. The validity of the derived rules is confirmed through applying them to the case of thin-angled monopole antenna and comparing the results with the measured. Finally using the spatial membership function context, input impedance of thick-angled monopole antenna is predicted and a novel view point to conventional electromagnetic parameters is discussed to generalize the modeling method.

We report on progress in the development of high current density NbN/AlN/NbN tunnel junctions for application as submillimeter wave SIS mixers. A ultra-high current density up to 120 kA/cm2, roughly two orders of magnitude larger than any reported results for all-NbN tunnel junctions, was achieved in the junctions. The magnetic field dependence and temperature dependence of critical supercurrents were measured to investigate the Josephson tunneling behaviour of critical supercurrents in the high-Jc junctions. We have developed a low-noise quasi-optical SIS mixer with the high-current density NbN/AlN/NbN junctions and two-junction tuning circuits which employ Al/SiO/NbN microstriplines. The tuning characteristics of the mixer were investigated by measuring the response in the direct detection mode by using the Fourier Transform Spectrometer (FTS) and measuring the response in the heterodyne detection mode with the standard Y-factor method at frequencies from 670 to 1082 GHz. An uncorrected double sideband receiver noise temperature of 457 K (12hν/kB) was obtained at 783 GHz.

We introduce an efficient interpolation attack which gives the tighter upper bound of the complexity and the number of pairs of plaintexts and ciphertexts required for the attack. In the previously known interpolation attack there is a problem in that the required complexity for the attack can be overestimated. We solve this problem by first, finding the actual number of coefficients in the polynomial used in the attack by using a computer algebra system, and second, by finding the polynomial with fewer coefficients by choosing the plaintexts. We apply this interpolation attack to the block cipher SNAKE and succeeded in attacking many ciphers in the SNAKE family. When we evaluate the resistance of a block cipher to interpolation attack, it is necessary to apply the interpolation attack described in this paper.

A support system for hepatectomy that allows segmentation of the liver interactively and directly on 3D images was developed. Intuitive 3D images of the liver and its vessels and tumors were drawn with an improved volume-rendering method. Regions supplied with blood by each branch were interactively identified. 3D segments were defined directly on the images using a mouse and excisions were estimated from these interactive inputs.

Methods to discover laws are reviewed from among both statistical approach and artificial intelligence approach with more emphasis placed on the latter. Dimensions discussed are variable dependency checking, passive or active data gathering, single or multiple laws discovery, static (equilibrium) or dynamic (transient) behavior, quantitative (numeric) or qualitative or structural law discovery, and use of domain-general knowledge. Some of the representative discovery systems are also briefly discussed in conjunction with the methods used in the above dimensions.

Interpolation attack was presented by Jakobsen and Knudsen at FSE'97. Interpolation attack is effective against ciphers that have a certain algebraic structure like the PURE cipher which is a prototype cipher, but it is difficult to apply the attack to real-world ciphers. This difficulty is due to the difficulty of deriving a low degree polynomial relation between ciphertexts and plaintexts. In other words, it is difficult to evaluate the security against interpolation attack. This paper generalizes the interpolation attack. The generalization makes easier to evaluate the security against interpolation attack. We call the generalized interpolation attack linear sum attack. We present an algorithm that evaluates the security of byte-oriented ciphers against linear sum attack. Moreover, we show the relationship between linear sum attack and higher order differential attack. In addition, we show the security of CRYPTON, E2, and RIJNDAEL against linear sum attack using the algorithm.

This paper describes the application of an unsupervised parallel approach called the Annealed Hopfield Neural Network (AHNN) using a modified cost function with moment and entropy preservation for magnetic resonance image (MRI) classification. In the AHNN, the neural network architecture is same as the original 2-D Hopfield net. And a new cooling schedule is embedded in order to make the modified energy function to converge to an equilibrium state. The idea is to formulate a clustering problem where the criterion for the optimum classification is chosen as the minimization of the Euclidean distance between training vectors and cluster-center vectors. In this article, the intensity of a pixel in an original image, the first moment combined with its neighbors, and their gray-level entropy are used to construct a 3-component training vector to map a neuron into a two-dimensional annealed Hopfield net. Although the simulated annealing method can yield the global minimum, it is very time-consuming with asymptotic iterations. In addition, to resolve the optimal problem using Hopfield or simulated annealing neural networks, the weighting factors to combine the penalty terms must be determined. The quality of final result is very sensitive to these weighting factors, and feasible values for them are difficult to find. Using the AHNN for magnetic resonance image classification, the need of finding weighting factors in the energy function can be eliminated and the rate of convergence is much faster than that of simulated annealing. The experimental results show that better and more valid solutions can be obtained using the AHNN than the previous approach in classification of the computer generated images. Promising solutions of MRI segmentation can be obtained using the proposed method. In addition, the convergence rates with different cooling schedules in the test phantom will be discussed.

A data-parallel processing approach is promising for real-time volume rendering because of the massive parallelism in volume rendering. In data-parallel volume rendering, local results processing elements(PEs) generate from allocated subvolumes are integrated to form a final image. Generally, the integration causes an overhead unavoidable in data-parallel volume rendering due to communications among PEs. This paper proposes a data-parallel shear-warp volume rendering algorithm combined with an adaptive volume subdivision method to reduce the communication overhead and improve processing efficiency. We implement the parallel algorithm on a message-passing multiprocessor system for performance evaluation. The experimental results show that the adaptive volume subdivision method can reduce the overhead and achieve higher efficiency compared with a conventional slab subdivision method.

Super-anomalous elliptic curves over a ring Z/nZ ;(n=Πi=1k piei) are defined by extending anomalous elliptic curves over a prime filed Fp. They have n points over a ring Z/nZ and pi points over Fpi for all pi. We generalize Satoh-Araki-Smart algorithm and Ruck algorithm, which solve a discrete logarithm problem over anomalous elliptic curves. We prove that a "discrete logarithm problem over super-anomalous elliptic curves" can be solved in deterministic polynomial time without knowing prime factors of n.

Watermarking methods that employ orthogonal transformations are very robust against non-geometrical modifications such as lossy compression, but attaining robustness against image translation or cropping is difficult. This report describes a watermarking method that increases robustness against geometrical modifications such as image translation and cropping by embedding watermark data in the frequency component of an image and detecting that data by considering the phase difference of the coefficients that results from translation of the image. Experimental results demonstrate the robustness of this method against both non-geometrical image changes and image translation and cropping.

This paper reports the design and characteristics of an aerial optical drop cable incorporating electric power wires, which was developed for a new π-system. The new system is called the power supply HUB π-system, in which commercial AC electric power is received at a central location of several optical network units (ONUs), and is distributed to each ONU by the aerial optical/electric drop cable. We describe the requirements for the cable, which guarantee a 20-year lifetime. We designed the cross-sectional structure of the cable, based on system requirements and operation requirements, and determined the strength wire type and diameter, based on the optical fiber failure prediction theory and a cable strain requirement. We confirmed that the cables, manufactured as a trial, have stable characteristics, which satisfy the above requirements. The optical/electric drop cables will be introduced in autumn 1999.

We report on the fabrication and operation of all-NbN single flux quantum (SFQ) circuits with resistively shunted NbN/AlN/NbN tunnel junctions fabricated on silicon substrates. The critical current varied by about 5% in 400 NbN/AlN/NbN junction arrays, where the junction area was 88 µm2. Critical current densities of the NbN/AlN/NbN tunnel junctions showed exponential dependence on the deposition time of the AlN barrier. By using the 12-nm-thick Cu film as shunted resistors, non-hysteretic current-voltage characteristics were achieved. From dc-SQUID measurements, the sheet inductance of our NbN stripline was estimated to be around 1.2 pH at 4.2 K. We designed and fabricated circuits consisting of dc/SFQ converters, Josephson transmission lines, and T flip-flop-based SFQ/dc converters. The circuits demonstrated correct operation with a bias margin of more than 15% at 4.2 K.

A multi-server system with trunk reservation is studied. The system is offered by two types of customers (class-1 and class-2). They arrive in independent batch Poisson streams and have an exponentially distributed service time. Class-1 customers will be lost or rejected if they find all S servers busy on their arrivals. Class-2 customers will use at most S'=S-R servers and enter a queue with N capacity if they find the number of idle servers less than or equal to R on their arrivals. Here, R is the number of reserved servers for class-1 customers. An example of the system is realized in NTT's facsimile communications network F-NET.

Measurement of the viscosity of liquid using a piezoelectric disk is described. Experiments with a radial expansion mode of a piezoceramic disk were carried out for water-glycerin mixture samples. Resonant resistance has linearity to the square root of the product of density and viscosity of a liquid around 113 kHz.

A multilayer perceptron is usually considered a passive learner that only receives given training data. However, if a multilayer perceptron actively gathers training data that resolve its uncertainty about a problem being learnt, sufficiently accurate classification is attained with fewer training data. Recently, such active learning has been receiving an increasing interest. In this paper, we propose a novel active learning strategy. The strategy attempts to produce only useful training data for multilayer perceptrons to achieve accurate classification, and avoids generating redundant training data. Furthermore, the strategy attempts to avoid generating temporarily useful training data that will become redundant in the future. As a result, the strategy can allow multilayer perceptrons to achieve accurate classification with fewer training data. To demonstrate the performance of the strategy in comparison with other active learning strategies, we also propose an empirical active learning algorithm as an implementation of the strategy, which does not require expensive computations. Experimental results show that the proposed algorithm improves the classification accuracy of a multilayer perceptron with fewer training data than that for a conventional random selection algorithm that constructs a training data set without explicit strategies. Moreover, the algorithm outperforms typical active learning algorithms in the experiments. Those results show that the algorithm can construct an appropriate training data set at lower computational cost, because training data generation is usually costly. Accordingly, the algorithm proves the effectiveness of the strategy through the experiments. We also discuss some drawbacks of the algorithm.