In this paper, we discuss computational methods for obtaining the bifurcation points and the branch directions at branching points of solution curves for the nonlinear resistive circuits. There are many kinds of the bifurcation points such as limit point, branch point and isolated point. At these points, the Jacobian matrix of circuit equation becomes singular so that we cannot directly apply the usual numerical techniques such as Newton-Raphson method. Therefore, we propose a simple modification technique such that the Newton-Raphson method can be also applied to the modified equations. On the other hand, a curve tracing algorithm can continuously trace the solution curves having the limit points and/or branching points. In this case, we can see whether the curve has passed through a bifurcation point or not by checking the sign of determinant of the Jacobian matrix. We also propose two different methods for calculating the directions of branches at branching point. Combining these algorithms, complicated solution curves will be easily traced by the curve tracing method. We show the example of a Hopfield network in Sect.5.

In the original shared buffer memory switch, there exists inherent unfairness since multicast cells always own higher priority to be served such that unicast cells must endure longer delay time. We propose a multicasting balancing method to overcome this unfair phenomenon in which the service order for two kinds of cells is decided by a ratio of unicast queue length and multicast queue length for each output port. Performance results are provided to verify the effectiveness.

We have proposed a connection-based optical wavelength division multiplexing network architecture. For the networks such as inter-office LANs, the guarantee of the bandwidths of connection-oriented calls is necessary. Notable features of the network are that multicast can be executed without copying the same data, and that time slots are rearrangeable to increase the throughput. The topology is passive star and a network controller (NWC) is connected to manage the time slot assignment. Each station's transmitting wavelength is fixed and is different from that of other stations. Each receiver changes the receiving wavelength slot by slot. Stations reserve time slots with permission of the NWC. Once a time slot is reserved the station can use the slot in every frame until the reservation is cancelled. This feature guarantees the bandwidths of connection-oriented calls. Upon receiving a time slot request, the NWC searches for a not-in-use slot common to the source station's transmitter (Tx) and the destination station's receiver (Rx). If there is no common empty slot and both the Tx and the Rx have empty slots, the NWC rearranges the already allocated time slots to create a new common empty slot. Simulations were performed to estimate the blocking rates for various cases of call bandwidth including multi-bitrate (the case in which various bandwidth calls are generated in a network) and multicast call, the calculation load of the NWC when it assigns a time slot including rearrangement, and the success rate of rearrangement. It was found that the blocking rate with the rearrangement is greatly reduced (1/10) compared with the case without rearrangement of the same throughput when the number of slots in a frame is more than 120, the number of stations in the network is 60 and the blocking rate without the rearrangement is less than 10-2. Over 100 Gbps throughput can be achieved when the number of slots in a frame is 120-240, the number of stations is 60, the bitrate of a transmitter is 2.5Gbps and the blocking rate is about 10-2. The rearrangement is especially effective in the case of multi-bitrate in which the blocking rate can be reduced to 1/100 that of the case without rearrangement at some point. It is also shown that a slot assignment including rearrangement can be executed sufficiently quickly (5s). These results indicate that practical realization of this access control architecture is possible.

The use of existing metallic local line facilities is being studied for providing "video on demand (VOD)" services to residential subscribers across asymmetric digital subscriber lines (ADSL). ADSL carries a high-rate channel in the downstream direction from a central office (CO) to the subscriber, and a low-rate channel in both directions on an existing 2-wire pair. Audio and video signals are compressed by the moving picture experts group's standardized algorithms (MPEG 1 and MPEG 2), and delivered to the subscriber in the high-rate channel. Control (demand and response) signals are transceived in the low-rate channel. This paper presents the line length coverage of ADSL systems given the environment of NTT's local networks. The bit rates in the downstream and upstream directions are assumed to be 1.6-9.2Mbit/s and 24kbit/s, respectively. Two types of ADSL systems are considered: transceiving ADSL signals using the plain old telephone service (POTS) line or the basic rate access (BRA; 320 kbaud ping-pong transmission system) line on the same 2-wire pair. 16-QAM, 32-QAM and 64-QAM are compared as transmission schemes. Intra-system crosstalk interference (interference between identical transmission systems) and inter-system crosstalk interference (interference between different transmission systems) with the existing digital subscriber lines (DSL) are estimated. It is shown that the inter-system crosstalk interference with BRA is most stringent, and ADSL with 16-QAM yields the best performance in NTT's local networks. This paper concludes that realizing ADSL with 16-QAM can achieve channel capacities of up to 9.2Mbit/s for fiber-in-the-feeder (FITF) access systems, but the possibility of applying ADSL to direct access systems is remote except for a restricted short haul use. Some comparisons regarding American local networks are also described.

This paper describes 950 GHz power performance of double-doped AlGaAs/InGaAs/AlGaAs heterojunction field-effect transistors (HJFET) operated at a drain bias voltage ranging from 2.5 to 3.5 V. The developed 1.0 µm gatelength HJFET exhibited a maximum drain current (Imax) of 500 mA/mm, a transconductance (gm) of 300 mS/mm, and a gate-to-drain breakdown voltage of 11 V. Operated at 3.0 V, a 17.5 mm gate periphery HJFET showed 1.4 W Pout and -50.3 dBc adjacent channel leakage power at a 50 kHz off-carrier frequency from 950 MHz with 50% PAE. Harmonic balance simulations revealed that the flat gm characteristics of the HJFET with respect to gate bias voltage are effective to suppress intermodulation distortion under large signal operation. The developed HJFET has great potential for small-sized digital cellular power applications operated at a low DC supply voltage.

We study all-to-all broadcasting in hypercubes with randomly distributed Byzantine faults. We construct an efficient broadcasting scheme BC1-n-cube running on the n-dimensional hypercube (n-cube for short) in 2n rounds, where for communication by each node of the n-cube, only one of its links is used in each round. The scheme BC1-n-cube can tolerate (n-1)/2 Byzantine faults of nodes and/or links in the worst case. If there are exactly f Byzantine faulty nodes randomly distributed in the n-cabe, BC1-n-cube succeeds with a probability higher than 1(64nf/2n) n/2. In other words, if 1/(64nk) of all the nodes(i.e., 2n/(64nk) nodes) fail in Byzantine manner randomly in the n-cube, then the scheme succeeds with a probability higher than 1kn/2. We also consider the case where all nodes are faultless but links may fail randomly in the n-cube. Broadcasting by BC1-n-cube is successful with a probability hig her than 1kn/2 provided that not more than 1/(64(n1)k) of all the links in the n-cube fail in Byzantine manner randomly. For the case where only links may fail, we give another broadcasting scheme BC2-n-cube which runs in 2n2 rounds. Broadcasting by BC2-n-cube is successful with a high probability if the number of Byzantine faulty links randomly distributed in the n-cube is not more than a constant fraction of the total number of links. That is, it succeeds with a probability higher than 1nkn/2 if 1/(48k) of all the links in the n-cube fail randomly in Byzantine manner.

This paper investigates the error correcting capabilities of concatenated codes employing algebraic geometry codes as outer codes and time-varying randomly selected inner codes, used on discrete memoryless channels with maximum likelihood decoding. It is proved that Gallager's random coding error exponent can be obtained for all rates by such codes. Further, it is clarified that the error exponent arbitrarily close to Gallager's can be obtained for almost all random selections of inner codes with a properly chosen code length, provided that the length of the outer code is sufficiently large. For a class of regular channels, the result is also valid for linear concatenated codes, and Gallager's expurgated error exponent can be asymptotically obtained for all rates.

This letter is concerned with the evaluation of the average computational complexity of the maximum likelihood decoding of a linear block code using its trellis diagram. Each section of the L-section minimal trellis diagram for a linear block code consists of parallel components which are structurally identical subgraphs without cross connection between them. A parallel component is also known to be decomposed into subgraphs, and a decoding algorithm by using the structure of the subgraphs of parallel components was proposed, and an upper bound on the worst case computational complexity was derived. In this letter, the average computational complexity of the decoding algorithm is evaluated by computer simulation. We evaluated the average numbers of additions and comparisons performed in the decoding algorithm for example codes, (64,45) extended and permuted binary primitive BCH code, the third order Reed-Muller (64,42) code and its (64,40) subcode. It is shown that the average numbers are much smaller than those for the worst case, and hence the decoding algorithm is efficient when the number of sections, L, is small, say 4 or 8, for the example codes. Especially, for the (64,45) extended binary primitive BCH code with L4, the average numbers of additions and comparisons in the decoding algorithm for finding the survivor's metric of each state after finding the smallest metric among parallel branches are about 1/50 and 17/100 of those in the conventional exhaustive search, respectively. The number of additions and comparisons by the conventional search for all the example codes is smallest when L is 4. As a result, the decoding algorithm with L4 gives the smallest number of operations among those decoding methods considered here.

Schnorr's identification scheme is the most efficient and simplest scheme based on the discrete logarithm problem. Unfortunately, Schnorr's scheme is not provably secure, i.e., the security has not been proven to be reducible to well defined intractable problems. Two works have already succeeded to construct provably secure variants of Schnorr's scheme. They have been constructed with a common approach, i.e., by modifying the formula to compute the public key so that each public key has multiple secret keys. These multiple secret keys seem to be essential for their provable security, but also give rise to a penalty in their efficiency. In this paper, we describe a new approach to constructing a provably secure variant, where we never modify the formula, and show that with our approach, we can construct a new efficient provably secure scheme.

This paper proposes a multiple optical wideband frequency modulation system and clarifies its phase noise insensitivity. In this system, an optical carrier is phase-modulated by a conventional FM signal to generate many sidebands in optical frequency band. The n-th order sideband component yields also FM signal with frequency deviation of n times the one of original FM signal. Therefore, by selecting the high order optical sideband, the wideband optical FM signal can be obtained. Moreover, if some sidebands are simultaneously extracted and multiplied at the receiver, a wideband FM signal with larger frequency deviation and no laser phase noise can be obtained, and FM threshold extension can be realized.

In this paper we propose a two-stage address coding scheme to transmit two data symbols at once within a frame in a MFSK/FH-CDMA system. We compare it with the conventional system using single-stage address coding. Assumed that the address codes of all users are known in the receiver. A multiuser detection scheme is applied and the performance is evaluated by computer simulations to show the improvement in bit error rate (BER) compairing to the conventional system. We also investigate the performance of error-correcting coding and decoding in the two-stage address coded MFSK/FH-CDMA system. An erasure decoding scheme is modified for the two-stage address coded system and is utilized to improve spectral efficiency or to increase user capacity in the MFSK/FH-CDMA system. Finally, we investigate a hybrid scheme of combining the multi-user detection scheme and the error-correcting decoding scheme for the two-stage address coded MFSK/FH-CDMA system. The performance is evaluated by computer simulations.

In this paper, we analyze the traffic data management requirements of the customers, describe the functions of the traffic database needed to satisfy their requirements, and propose a highly distributed database system which can efficiently implement these functions. Finally, we report the results of system performance evaluations.

This paper proposes a new class of error locating codes which corrects random single-bit errors and indicates a location of an erroneous b-bit byte which includes e-bit errors, where 2 e b, called SECSe/bEL codes. This type of codes is very suitable for an application to memory systems constructed from byte-organized memory chips because this corrects random single-bit errors induced by soft-errors and also indicates the position of the faulty memory chips. This paper also gives a construction method of the proposed codes using tensor product of the two codes, i.e., the single b-bit byte error correcting codes and the single-bit error correcting and e-bit error detecting codes. This clarifies lower bounds and error control capabilities of the proposed codes.

This paper is concerned with the evaluation of the block error probability Pic of a block modulation code for closest coset decoding over an AWGN channel. In most cases, the conventional union bound on Pic for closest coset decoding is loose not only at low signal-to-noise ratios but at relatively high signal-to-noise ratios. In this paper, we introduce a new upper bound on the probability of union of events by using the graph theory and we derive an improved upper bound on Pic for some block modulation codes using closest coset decoding over an AWGN channel. We show that the new bound is better than the conventional union bound especially at relatively high signal-to-noise ratios.

When bit error probability of a trellis-coded modulation (TCM) scheme becomes very small, it is almost impossible to evaluate it by an ordinary Monte-Carlo simulation method. Importance sampling is a technique of reducing the number of simulation samples required. The reduction is attained by modifying the noise to produce more errors. The low error rate can be effectively estimated by applying importance sampling. Each simulation run simulates a single error event, and importance sampling is used to make the error events more frequent. The previous design method of the probability density function in importance sampling is not suitable for the TCM scheme on an additive non-Gaussian noise channel. The main problem is how to design the probability density function of the noise used in the simulation. We propose a new design method of the simulation probability density function related to the Bhattacharyya bound. It is reduced to the same simulation probability density function of the old method when the noise is additive white Gaussian. By using the proposed method for an additive non-Gaussian noise, the reduction of simulation time is about 1/170 at bit error rate of 106 if the overhead of the calculation of the Bhattacharyya bound is ignored. Under the same condition, the reduction of the simulation time by the proposed method is 1/65 of the ordinary Monte-Carlo method even if we take the overhead for importance sampling into account.

A data-base for data compression is universal if in its construction no prior knowledge of the source distribution is assumed and is optimal if, when we encode the reference index of the data-base, its encoding rate achieves the optimal encoding rate for any given source: in the noiseless case the entropy rate and in the semifaithful case the rate-distortion function of the source. In the present paper, we construct a universal data-base for all stationary ergodic sources, and prove the optimality of the thus constructed data-base for two typical methods of referring to the data-base: one is a block-shift type reference and the other is a single-shift type reference.

The periodic correlation properties of M sequences coded by channel codes are discussed. As for the channel codes, the Manchester code and the eight DC free codes in the FM family codes, which include the conventional FM code and the differential Manchester code, are adopted. The M sequences coded by the DC free codes in the FM family codes are referred to as FM coded M sequences. The periodic correlation properties of all combinations of the FM coded M sequences are checked, and the combinations which can provide almost the same or better properties as compared with those of the preferred pairs of M sequences are described. An example of code design using the FM coded M sequences for asynchronous direct sequence/spread spectrum multiple access systems is also discussed.

We have developed a 15-Gbit/s 96-gate Si-bipolar gate array using 0.5-µm Si-bipolar technology, a sophisticated internal cell design, an I/O buffer design suitable for high-speed operation and high-frequency package technology. The decision circuit and 4 : 1 multiplexer fabricated on the gate array operate up to 15-Gbit/s and above 10-Gbit/s respectively. The data input sensitivity and the phase margin of the decision circuit are 53 mVpp and 288 at 10-Gbit/s operation. This gate array promises to be useful in shortening the development period and lowering cost of 10-Gbit/s class IC's.

Silver is a fundamental material for electrical contact application. In spite of high electrical conductivity and economical advantage, silver surface is corroded easily by environment contained sulfide. A corrosion product as Ag2S deteriorates the property of contact reliability. In order to examine contact reliability, the acceleration tests have been accepted widely in industries. In the present study, the acceleration factor of the contact reliability for the sulfide film on the surface of silver contact which was subject to the tarnish acceleration test was clarified in comparison with the film grown in a normal office environment. The accelerated environment based on the Japan Electric Industry Development Association (JEIDA) standard No.25 was adopted. This environment is consisted of air contained 3 ppm H2S gas under 40, 85-95% RH. The growth rate of the sulfide film (Ag2S) was evaluated by applying the ellipsometry analysis. In the results, it was found that growth of Ag2S film of 500 in thickness in the normal office environment required corrosion time of 3103 h. This thickness of 500 caused increase in contact resistance of 0.1-1.0 (Ω). However, in the accelerated environment, corrosion time decreased to 1.7 h for same thickness. Therefore, the acceleration factor was obtained by comparison of these time as 1.8103 for the standard test of JEIDA.

This paper investigates the effect of crosstalk in multichannel coherent optical ASK systems. A closed-form signal-to-noise ratio (SNR) expression as a function of number of channels, channel separation, laser phase noise, intermediate frequency (IF) filter bandwidth expansion factor (α), system bit rate (Rb=1/T), and additive shot noise is presented. When the desired channel is between two channels in the electrical domain, the minimum permissible electrical domain channel spacing for a 1dB sensitivity penalty due to crosstalk is found to be 4.85Rb when α is optimum; and 8Rb when α=5 for νT=0.30 at BER=10-9. A fairly good agreement is found between the results of this work and those of a previous study.