This paper describes the requirements for fault recovery on photonic networks and proposes a fast restoration scheme for recovering optical networks. The proposed scheme is a type of pre-assignment restoration. The features of the scheme are that it is suitable for multi-recovery classes aimed at fine control of the optical paths and that it establishes harmonization between restoration control and distributed network control such as in IP networks. The scheme is implemented on Photonic multi protocol label switching (MPLS) routers. A restoration demonstration was performed and recovery was achieved within 500ms in the optical layer.

High-speed learning of neural networks using the multifrequency oscillation method is demonstrated for first time. During the learning of an analog neural network integrated circuit implementing the exclusive-OR' logic, weight and threshold values converge to steady states within 2 ms for a learning speed of 2 mega-patterns per second.

This paper discusses global area optical transport ring networks using wavelength division multiplexing (WDM) technologies and proposes a novel optical add/drop multiplexer (OADM) architecture suitable for such an application field. Study on the requirements of a global area ring application elucidates the appropriate ring/protection architecture as the path switched bi-directional ring. The proposed OADM architecture has flexibility in terms of path provisioning and scalability. We conclude that the proposed OADM can effectively configure the large-scale path switched bi-directional rings.

This paper discusses global area optical transport ring networks using wavelength division multiplexing (WDM) technologies and proposes a novel optical add/drop multiplexer (OADM) architecture suitable for such an application field. Study on the requirements of a global area ring application elucidates the appropriate ring/protection architecture as the path switched bi-directional ring. The proposed OADM architecture has flexibility in terms of path provisioning and scalability. We conclude that the proposed OADM can effectively configure the large-scale path switched bi-directional rings.

In order that they fully support human activities, new network services and applications are overwhelming conventional ones, such as telephony, facsimile, and telegraph. Demands for digital networks are exploding, not only in terms of quantity but also quality. Nobody can predict where these demands will lead. Traffic engineering, which is impossible in pure Internet protocol (IP) -based networks, is recognized as being indispensable for quality of service (QoS) control. It includes guaranteed services in terms of bandwidth, delay, delay variation (jitter), and service protection. The "engineered tunnel" through IP network supports virtual private networks (VPNs) and allows us to develop voice-over-IP (VoIP), teleconferencing and other secure private network services. This paper proposes the "photonic router" which makes use of wavelength-based networks for signal routing. IP packets having the same destination are bundled into a wavelength path. Interchange nodes along the path route control path routing on the basis of wavelength information, not on IP headers, which can not be read or processed with current optical techniques. In short, wavelength path routing offers "cut-through" in the photonic layer. This paper shows its feasibility by describing the combination of an optical cross-connect, payload assembler/disassembler, label controller, and IP router. Optical cross-connect systems, which are now being intensively studied worldwide, are deemed to be key equipment for a wavelength-path network with centralized control system. This paper proposes to apply the cross-connect to an IP network with distributed autonomous control.

Three-dimensional MEMS optical switches(3D-MEMS) and planar lightwave circuit thermo-optical switches (PLC-TOSW) are suitable for use in large-scale photonic cross-connect switches (PXCs). Usually, such large-scale optical switches are created by integrating many small switch elements (such as 12 switches). Therefore, the reliability or lifetime of the individual switch element greatly affects the reliability of the PXC system. In this paper, the effect of the number and failure probability of switch elements on PXC reliability is statistically estimated. First, the equivalent number of switch elements needed to compose a large-scale PXC is determined for the switch types mentioned above. Based on this evaluation, yields and lifetimes of switch modules are estimated for each switch type. The improvement in reliability due to switch duplication (preparing spare switches) is also estimated and discussed.

A configuration of multi/demultiplexer using a 4-port optical circulator and interference filters is presented. Owing to the circulator, interference filters for the signal separation can be set normal to the optical signal propagating direction. Demultiplexing operation for 3 wavelength-multiplexed signals, of which wavelength spacings are 1.8 and 4.9 nm, is experimentally confirmed.

This paper investigates approaches that can cancel nonlinear phase noise effectively for the phase-conjugate pair diversity transmission of 16-QAM WDM signals through multi-core fiber. The geometric mean is introduced for the combination of the phase-conjugate pair. A numerical simulation suggests that span-by-span chromatic dispersion compensation is more effective at cancelling phase noise in long distance transmission than lumped compensation at the receiver. Simulations suggest the span-wise compensation described herein yields Q-value enhancement of 7.8 and 6.8dB for CD values of 10 and 20.6ps/nm/km, respectively, whereas the lumped compensation equivalent attains only 3.5dB. A 1050km recirculating loop experiment confirmed a Q-value enhancement of 4.1dB for 20.6ps/nm/km, span-wise compensation transmission.

A very-high-speed ten-neuron analog neural network LSI chip is fabricated for the first time using super self-aligned Si bipolar process technology. The LSI consists of ten neurons and 100 electrically modifiable synaptic weights. The neural network nonlinear mapping function to solve the four-bit parity problem is successfully demonstrated at 150 mega-patterns/sec. The operation speed of this neural network is, to the best of the authors, knowledge, the fastest yet reported.

This paper describes the recently developed 816 delivery and coupling switch (DC-switch) boards for constructing a 320-Gb/s throughput (2. 5 Gb/s 8 multiplexed wavelengths 16 incoming/outgoing link pairs) optical path cross-connect (OPXC) system based on wavelength path (WP) and virtual wavelength path (VWP) schemes. The DC-switch-based OPXC system, compared with conventional space division switch (SD-switch)-based OPXC system architecture, is shown to be superior in terms of; i) high link modularity, ii) upgradability from WP network to VWP network, iii) better transmission characteristics, and iv) lower total switching power consumption. Therefore, the DC-switch-based OPXC system can realize cost-effective optical path networks. The developed DC-switches exploit the silica-based planar lightwave circuit (PLC) technologies, and DC-switch board size is 300330 mm2 (one switch). The worst values of the insertion loss of the board, ON/OFF ratio, and polarization dependent loss are 14. 5 dB, 34 dB and 0. 5 dB, respectively. Moreover, even though switching is realized by thermo-optic effects, the optical output level varies by only 0. 7 dB and 0. 8 dB for ON- and OFF-state signals, respectively, when the environmental temperature is varied from 5 to 65 .