Banyan networks are used in multiprocessor computer applications for an ATM switching. In this paper, we study the continuous blocking of the first n-stage which makes the performance of the banyan networks decrease. We use the 2-dilated banyan networks into the banyan networks to remove the continuous blocking of the first n-stage. We call the new networks as the hybrid dilated banyan networks. We explain how to analyze the throughput of this networks at each stage. Based on the analysis of input rate and output rate at each stage, we can design the hybrid dilated banyan networks with the desirable output rate. The result of analysis shows the hybrid dilated banyan networks have higher performance and feasibility than the banyan networks.

Recurrent neural networks have the potential of performing parallel computation for associative memory and optimization, which is realized by the electronic implementation of neural networks in VLSI technology. Since the time delays in real electronic implementation of neural networks are unavoidably encountered and they can cause systems to oscillate, it is thus practically important to investigate the qualitative properties of neural networks with time delays. In this paper, a class of sufficient conditions is obtained, under which neural networks are globally asymptotically stable independent of time delays.

Hankel's asymptotic expansions are frequently used for numerical calculation of cylindrical functions of complex order. We beforehand need to estimate the precisions of the cylindrical functions calculated with Hankel's asymptotic expansions in order to use these expansions. This letter presents comparatively simple expressions for rough estimations of the errors of the cylindrical functions calculated with the asymptotic expansions, and features of the errors are discussed.

In this paper, we propose a new transaction numbering scheme and a new validation scheme for controlling transactions optimistically in client-server architectural mobile distributed database systems (MDDBSs). In the system, mobile units (MUs) request transaction-related services, e.g., concurrency control, commit process, then the mobile support stations (MSSs) provide the required services. The mobile computing environment makes it very difficult for each MU to assign unique transaction number to transactions since it is allowed to move in communication disconnected states. Besides, validating transactions numbered by the previous transaction numbering scheme could wait indefinitely in the case of data transfer delay. Thus, we propose a new transaction numbering scheme called datatransfer time oriented transaction numbering scheme (DATTO) ,in which we can remove waiting time for validation by determining validation-start time with data-transfer completion time. However, if the previous validation scheme for the static environment is directly applied transactions numbered by DATTO, undesirable results could occur in abnormal cases due to latency on the wireless communication. Thus, we also propose a new validation scheme, called commit-order oriented validation (COOV) ,which is always able to produce correct results by applying backward validation to the abnormal cases.

As the demand for communications via satellite is rapidly increasing, techniques that produce large traffic capacity are becoming more and more appreciated. We present a class of block coded modulation (BCM) and multiple block coded modulation (MBCM) schemes in this paper. While the BCM scheme is directly derived from our previous work, the MBCM schemes are newly developed using a technique of multiple symbol transmission via a single trellis branch. This class of BCM and MBCM schemes is both power and bandwidth efficient. They also have an advantage in holding both a trellis and a block structure. Code structures, decoding trellises and the corresponding branch variables of these BCM and MBCM schemes are all derived. Their applications to satellite communications are discussed. Computer simulations are performed to verify coding gain performance.

This paper proposes a new AFC (automatic frequency control) circuit employing a double-product type frequency discriminator to enable fast acquisition in very-low CNR (carrier to noise power ratio) environments. The frequency step responses of the proposed AFC circuit are theoretically analyzed. In addition this paper evaluates the performance of the proposed AFC circuit by computer simulation in very-low CNR environments. The simulation results confirm that click noise at the frequency discriminator causes large frequency tracking error and that this error can be improved by increasing the delay time of the double-product type frequency discriminator. The frequency error can be also reduced by introducing the proposed frequency discriminator to modify the frequency error detection performance. The acquisition time of the proposed AFC circuit can be reduced by about 100 symbols compared to the conventional cross-product type AFC circuit.

A new simple cell spacing architecture that guarantees the peak cell interval and realizes preferential contention resolution is proposed. Scheduling the cell emission on departure of the previous cell, not arrival, allows the source peak cell interval to be regenerated without clumping. Priority control is also realized in the proposed spacer. A connection is scheduled either at the head or tail of the contention chain depending on its priority. The proposed method is applied to realize the UPC function. The proposed cell spacer eliminates the clumping effects of CDV completely and achieves high bandwidth efficiency.

A new learning algorithm is proposed to enhance fault tolerance ability of the feedforward neural networks. The algorithm focuses on the links (weights) that may cause errors at the output when they are open faults. The relevances of the synaptic weights to the output error (i.e. the sensitivity of the output error to the weight fault) are estimated in each training cycle of the standard backpropagation using the Taylor expansion of the output around fault-free weights. Then the weight giving the maximum relevance is decreased. The approach taken by the algorithm described in this paper is to prevent the weights from having large relevances. The simulation results indicate that the network trained with the proposed algorithm do have significantly better fault tolerance than the network trained with the standard backpropagation algorithm. The simulation results show that the fault tolerance and the generalization abilities are improved.

We have investigated the BER performance of TC 8PSK with 2 branch SC and MRC diversities on spatially correlated Rayleigh fading channel. The upper bounds using the transfer function bounding technique are derived several numerical results are shown. Although the correlation between branches causes signal-to-noise (SNR) loss (relative to uncorrelated fading case) for SC and MRC diversities, the diversity can lead to achieve the diversity gain compared to the system without diversity. It is found that the diversity gain of 4-state TC 8PSK is larger than 8-state TC 8PSK. It is also shown that the BER performance of TC 8PSK is decreased as the antenna separation is decreased.

In this paper, we investigate statistical and perfect Knowledge Complexity (KC) with respect ot oracle entropy and average case oracle measures. As main results. we show the following: (1) for any k(n) 1/poly (n), if a language L has perfect KC k(n) + n-ω(1) with respect to oracle entropy measure, then L has perfect KC k(n) with respect to oracle entropy measure (Theorem 3.1); (2) for any k(n) 1/poly(n), if a language L has perfect KC k(n) + n-ω(1) with respect to average case oracle measure, then L has perfect KC k(n) with respect to average case oracle measure (Theorem 3.2); (3) if a language L has statistical KC k(n) ο(1) with respect to oracle entropy measure, then for any ε > 0, L has statistical KC k(n) + 1 + ε with respect to average case oracle measure (Theorem 4.1); and (4) if a language L has perfect KC k(n) ο(1) with respect to oracle entropy measure, then for any ε > 0, L has perfect KC k(n) + 2 + ε with respect to average case oracle measure (Theorem 4.2).

We propose an InGaAlAs waveguide p-i-n photodiode (WG-PD) with a thick symmetric double-core for surface-hybrid integration onto optical platforms, which can be applied to low cost optical modules for access networks. The waveguide structure is designed to efficiently couple to flat-ended single mode fibers while maintaining low-voltage (less than 2 V) operation. Crystal growth conditions and a passivation technique are also investigated for obtaining high responsivity, low dark current and highly reliable operation. Fiber-coupled responsivity as high as 0.95 A/W, at a 1.3-µm wavelength, and vertical coupling tolerance as wide as 2.6 µm are demonstrated for a dispersion-shifted fiber (DSF) coupling at an operating voltage of 2 V. Dark current is as low as 300 pA at 25 and 12 nA at 100. A temperature accelerated aging test is performed to show the feasibility of using the WG-PD in long-term practical applications.

In-line transceiver chip emitting at 1.3 µm and receiving at 1.55 µm is described regarding the crosstalk occurring between the 1.3-µm laser and the 1.55-µm integrated photodiode. Contribution of optical and electrical crosstalk to the overall crosstalk is measured and discussed. Techniques to overcome the crosstalk are proposed and demonstrated by showing the feasibility of system compatible diplex sensitivities.

Polarization insensitive discrete electroabsorption modulators have been designed as an optical gating device. It reveals the first finding, to our knowledge, that the ratio of the optical confinement factor (Γ) to the differential of the values (ΔΓ) between TE and TM polarized lights decides polarization dependence of attenuation. The ratio ΔΓ/Γ is significantly reduced by increasing core thickness. Large optical confinement structures combining a thick InGaAsP bulk absorption layer and polyimide-buried mesa-ridge waveguide have fabricated. The ratio ΔΓ/Γ of the high-mesa structure was estimated to be less than 0.05 in the gain-region of an erbium-doped fiber amplifier (EDFA), which enable us extremely low polarization sensitivity less than 1 dB up to 20 dB extinction. Proper waveguide length of the structure allowed low insertion loss ( 9.3 dB), small loss-change ( 1.8 dB) and sufficient modulation depth ( 30 dB) simultaneously in the EDFA's gain region. The low-mesa structure provided low insertion loss around 7 dB with small deviation in the wavelength region. High modulation band-width and a polarization-insensitive optical gating waveform have also demonstrated.

This paper describes a novel quasi-transmission-line variable-reactance circuit that extends the variable-phase range of phase shifters. It consists of a transmission line and two shunt varactors. By appropriately choosing the characteristic impedance and electrical length of the transmission line, the variable-phase range can be significantly increased. Since the proposed circuit can be fabricated by the conventional MESFET process, a phase shifter can be integrated with other functional circuits. This enables fully monolithic integration of RF circuits as a one-chip multi-functional MMIC in radio communication systems. The variable-phase range of the prototype X-band monolithic phase shifter is 208 degrees, which is approximately four times as large as that of conventional one.

Micromachined sensors and actuators applied with electrostatic fields are getting widely developed. At the same time, "electrets," which are dielectrics carrying non-equilibrium permanent space charges or polarization distribution, are in demand because they improve the transducer characteristics. In this paper, we have reported on our successful fabrication of silicon dioxide electrets with extremely superior long-term charge stability by plasma chemical vapor deposition (PCVD). We have also reported on the correlation between the deposition conditions, the long-term charge stability and thermally stimulated current (TSC). Finally, the characterization of the long-term stable electrets will be described and discussed.

It has been known that an N-vertex binary tree can be embedded into the path and grid with dilation O(N/logN) and O((N/logN)), respectively. This paper shows that an N-vertex binary tree with proper pathwidth at most k can be embedded into the path grid with dilation O(N/N1/k) and O((N/N1/2k)), respectively.

Nyberg and Knudsen proved that the maximum average of differential probability (ADPmax) and the maximum average of linear probability (ALPmax) of Feistel cipher with over 4 rounds can be evaluated as ADPmax 2DCP2max and ALPmax 2LCP2max using the maximum of defferential characteristic probability (DCPmax) and the maximum of linear characteristic probability (LCPmax) per round. This paper shows ADPmax DCP2max and ALPmax LCP2max if the F function is a bijection and the Feistel cipher has more than 3 rounds. The results prove that Feistel ciphers are stronger against differential and linear cryptanalyses than previously thought. Combining this result with that of Luby and Rackoff, the implication is that the 3-round Feistel cipher could be used as a building block cipher for the construction of provable secure block cipher algorithm.

In estimating the vulnerability of a block cipher to differential cryptanalysis and linear cryptanalysis, we must consider the fact that the differential probability and the linear probability vary with the key. In the case of cryptosystems where the round key is XORed to the input data of each round, the difference in both types of probability with different keys is regarded as negligible. However, this is not the case with RC5. This paper makes a primary analysis of the key-dependency of linear probability of RC5. Throughout this paper we study "precise" linear probability. We find some linear approximations that have higher deviation (bias) for some keys than the "best linear approximation" claimed by Kaliski and Yin in CRYPTO'95. Using one linear approximation, we find 10 weak keys of RC5-4/2/2 with linear probability 2-1, 2 weak keys of RC5-4/5/16 with linear probability 2-2, and a weak key of RC5-16/5/16 with linear probability 2-15.4, while Kaliski-Yin's "best biases" are 2-3, 2-9, and 2-17, respectively.

We review a fundamental weak point of the OSS digital signature scheme against cryptanalysis by Pollard et al., and propose a new scheme of digital signature which overcomes this defect. More specifically, instead of the ring of the rational integer, we use the ring of integral quaternions, which is a non-commutative Euclidean ring. Known attacks to OSS signature do not work our scheme due to the non-commutativity. On the other hand, this scheme causes little increase in the burden of generation and verification of digital signature for the legitimate users, with respect to the original OSS scheme.

In this paper a new type of public-key cryptosystem, proxy cryptosystem, is studied. The proxy cryptosystem allows an original decryptor to transform its ciphertext to a ciphertext for a designated decryptor, proxy decryptor. Once the ciphertext transformation is executed, the proxy decryptor can compute a plaintext in place of the original decryptor. Such a cryptosystem is very useful when an entity has to deal with large amount of decrypting operation. The entity can actually speed-up the decrypting operation by authorizing multiple proxy decyptors. Concrete proxy cryptosystems are constructed for the ElGamal cryptosystem and the RSA cryptosystem. A straightforward construction of the proxy cryptosystem is given as follows. The original decryptor decrypts its ciphertext and re-encrypts an obtained plaintext under a designated proxy decryptor's public key. Then the designated proxy decryptor can read the plaintext. Our constructions are more efficient than such consecutive execution of decryption and re-encryption. Especially, the computational work done by the original decryptor is reduced in the proxy cryptosystems.