In optical networks using wavelength routing the ability to assign wavelength channels independently on each link is very useful but to maintain optical transparency then becomes very difficult. In fact optical transparency is not a specific requirement of networks that transmit signals independent of their format. Transponders can provide wavelength conversion if adequate performance can be achieved in the electronic routing of signals independent of format. We demonstrate experimentally a 10 multiplied by 10 matrix switch that is bit-rate, format, and modulation transparent for analogue signals and data at various rates up to 2.5 Gbps. The switch architecture is suitable for use in a wavelength converting transponder node. Prospects for staging these devices at OC-48 and OC-192 rates are discussed.

Polarization independent, high speed tunable (approximately ns) optical filters/receivers are key components in future broadband switching networks. In this paper, the theory and experiment of a novel polarization independent tunable filter based on grating-assisted InGaAsP/InP directional coupler are reviewed. Such a filter, made of three planar coupled hetero-material waveguides, has the advantages of yield compensation after fabrication and low cross-talks between the TE and TM modes. It can also be broadly tuned while still maintaining polarization-independent performance. The approaches toward ultra-low sidelobe filter as well as 'window-'like filter spectrum also are discussed.

Tunable optical filters are important building blocks for all-optical systems and networks. Fast optical tuning in several microseconds is necessary to perform high-speed optical packet switching. Multi-gigabit/sec packet-switching will provide flexibility and higher network throughput when large numbers of users communicate simultaneously. One approach to achieve fast wavelength tuning is to use high- speed piezoelectrically driven fiber Fabry-Perot tunable filters (FFP-TFs). The requirement for tuning in microseconds raises a whole new set of challenges, such as ringing, thermostability and mechanical inertia control. It was shown that correlation between the mechanical resonance and optical response of the filter is important for the filter's speed and for mounting hardware and control circuitry optimization. These features together with the FFP-TF's high capacitance (approximately 0.25 - 0.5 microfarad) are being folded into building a special controller to substantially improve the shape of the driving signal and the response of the filter. The resultant controller enables tuning the high-speed FFP-TF three- orders-of-magnitude faster than that possible with standard commercial FFP-TFs. The fastest switching time achieved is 2.5 microseconds. As the result, a new packet-switched media access control protocol is being designed to minimize the searching time. The filter scans only once through the entire optical region and then tunes to all the required channels one after another in a few microseconds. It can help update Rainbow-2 Broadcast-and-Select High-Speed Wavelength Division Multiplexing All-Optical network that currently has a circuit-switched protocol using standard FFP-TFs.

The steady-state soliton parameters are studied both for anomalous and normal dispersion regimes of a fiber laser, taking into account the effects of fiber dispersion and nonlinearity, linear gain and gain dispersion, nonlinear gain and two-photon absorption. It is shown that both the soliton chirp and width are significantly larger for normal than for anomalous dispersion. The existence of singularities for the soliton width and amplitude are revealed for both dispersion regimes. In the case of anomalous dispersion, the soliton perturbation theory and the phase-plane formalism are used to investigate the stability of the steady-state solution.

The mutual interaction between two solitons, frequency shifted from each other, and copropagating in a non-linear media, is investigated for the case of a wide range of frequency shifts. Analytical solutions are derived for the trajectory of solitons when submitted to frequency shifts. Simulations are also presented as certification of the validity limits of each analytical result. Numerical results are used to define the optimum range of frequency shifts, and potential applications are proposed.

Microwave coplanar waveguide slow wave structures, suitable for traveling wave electro-optic modulators, were experimentally investigated up to 40 GHz. Velocity matching was achieved by introducing different capacitive elements between the signal and the ground electrodes of otherwise uniform coplanar waveguides (CPWs). Four different types of slow wave structures and uniform coplanar waveguides were designed and fabricated on both semi-insulating GaAs (SI- GaAs) and semi-insulating InP (SI-InP) substrates. The measured s-parameters were used to calculate the characteristic impedance, phase velocity, and microwave loss of these structures. The results are in close agreement with the design values and demonstrate an improved performance on SI-InP substrate.

The small-signal microwave performance of 1.3 micrometer gain-coupled DFB lasers has been measured as a function of device parameters such as current and ridge width by treating the laser as a two port device. The s-parameters were measured over a frequency range from 0.5 GHz to 8 GHz. The simulation software LIBRA was used with S₁₁ return loss measurements to extract equivalent circuit parameters according to Kan and Lau's laser model. Results demonstrate that the parameters describing the input impedance of the laser undergo a discontinuity at the lasing threshold current, I_th. The measured optical intensity modulation (IM) response, (S₂₁), was used to study the behavior of the relaxation oscillation frequency f_r as a function of laser bias current. It has been demonstrated that at high laser bias current I_laser, f_r is independent of laser bias current (I_laser). We have also used the equivalent circuit parameters obtained from the S₁₁ data to predict values for f_r based on equations derived. These were then compared to the f_r values obtained from S₂₁ measurements.

A highly angular-dispersive element called the virtually- imaged phased-array (VIPA) is reviewed. The fundamental operation of the VIPA and its application to a wavelength demultiplexer are discussed. The improvements both in the filtering bandwidth and in the temperature dependence are also described.

Concave gratings traditionally have been based on the Rowland circle design. A grating design formalism is discussed here that allows greater flexibility than the Rowland circle design. With this formalism, a wavelength division demultiplexer (WDM) may be designed so that the passbands for the output waveguides may be tuned after the demultiplexer has been fabricated. The design considerations for these devices are discussed and are different from the design considerations for Rowland circle gratings. The design of the grating facets is discussed in detail. Experimental results show that facets using light at an angle of incidence that is nominally 45 degrees can become highly efficient due to total internal reflection.

Based on the analysis of the design theories for non- collinear AOTF with consideration of the birefringence and the rotatory property, the theory of imaging resolution for non-collinear AOTF and the factors which related to imaging resolution were studied. The dependences of diffracted optical beam spread with bandpass (Delta) (lambda) as well as d (theta) _d/d(lambda) with incident wavelength (lambda) _o and incident angle (theta) _i were calculated. The results obtained prove a basis for improving the design of AOTF with large aperture.

We demonstrate near-error-free output-port contention resolution in a WDM switching node based on all-optical wavelength shifting. Contention between incoming 1-Gbit/s data packets is resolved in real-time by comparing subcarrier-multiplexed control headers on 2 input ports and then all-optically wavelength-shifting one input signal to a free wavelength if contention occurs. Additionally, the 50 Mbit/s subcarrier header itself is replaced and updated with new routing information.

System design and performance for reconfigurable and simultaneous 2-D multiple-plane WDM optical interconnects has been investigated. Device characteristics such as detector responsivity shape, efficiency, and complexity impact design requirements on wavelength separation and channel power. For reconfigurable interconnects, signal power penalties impact system design whereas for simultaneous interconnects, power penalties as well as optical crosstalk from the detection of undesired channels impact system design.

The performance of a hybrid AM-VSB/BPSK optical fiber transmission system utilizing a 1.55 micrometer distributed feedback (DFB) laser and an Er doped optical amplifier is presented. A BPSK modulated 2 MHz pseudo-random digital channel is substituted for one of the AM channels in a 60 channel CATV system and optically transmitted using a directly modulated 1.55 micrometer DFB laser. The intermodulation distortion effects with and without optical amplification were studied. No degradation of the AM channels is observed unless the modulation depth of the BPSK channel is increased to the point where laser clipping effects become significant.

This paper describes a new structure of photonic add/drop multiplexer (ADM), called optical wavelength ADM (OW-ADM), discusses the applications in multi-wavelength optical ring networks and the network characters. This scheme uses optical wavelength as addressing information of data channel and performs all-optical add/drop function at each node by sensing the wavelength, without converting the high-speed data to electronic form. The key components in OW-ADM include wavelength grating router (WGR), optical switch, optical coupler and band reject filter. The thought is based on wavelength sequence selection (WSS) property of WGR. This OW-ADM can avoid the bottlenecks associated with the convention between optical signal and electrical signal and accomplish all-optical transparent transmission. It also has the following good features: (1) all-optical self-healing; (2) dual-function of ADM and protection switching; (3) self- test function; (4) small crosstalk and small wavelength interval; (5) wavelength-reuse capability, etc.

We address the problem of scheduling data bursts with no intermediate queuing in high-speed networks. The problem is to determine the exact intermediate switch schedules such that each burst travels from source to destination without any queuing. We target applications which require high-speed burst transmissions and which can tolerate some degree of latency between requesting transmissions and the actual transmission of the burst itself. Such scenarios are common in satellite networks where the characteristics of the satellite links (multiple access, long delays) and the bursty nature of the traffic make it difficult to apply traditional network algorithms. In this paper, we analyze reservation-based algorithms for just-in-time scheduling of data bursts. These algorithms have been developed as part of the highball project, a high speed, packet-switched network using distributed reservation and scheduling algorithms. Simulations of these algorithms on various network topologies have shown good scheduling efficiencies. However, the data bursts incur a large scheduling delay since each reservation request must be received by all the nodes in the network. We then describe an improved scheduling algorithm which achieves higher scheduling efficiencies and lower scheduling delays by determining the schedule before all the nodes have received the request. We prove theoretically that the resulting burst schedules are consistent and collision- free, and analyze the performance of the algorithm via simulation. This idea can also be adapted to other parallel discrete event simulation (PDES) applications, such as intelligent highways, traffic control and distributed interactive simulation.

The development of a simulation model for selective-area epitaxy and the fabrication of semiconductor lasers monolithically integrated with electroabsorption modulators by this technique are presented. Diffusion equations and boundary conditions from selective-area MOCVD theory are applied in a computational model to predict column III reactant concentrations, and self-consistent solutions for reaction parameters are found using the finite element method. Data are presented to demonstrate accurate predictions of the thickness and composition of selectively grown ternary InGaAs quantum wells. This model was utilized to design the selective growth mask for Fabry-Perot lasers integrated with intracavity electroabsorption modulators. These devices, with modulator lengths of 290, 620, and 1020 micrometer, exhibit cw threshold currents of 9, 7.5, and 7.5 mA, respectively. Also, extinction ratios of 16.5, 19.5, and 20.5 dB, respectively, are measured at a modulator reverse bias of 2 V. Distributed Bragg reflector lasers with monolithically integrated external cavity modulators are also fabricated, and the selective-area MOCVD simulation was employed to design the growth mask dimensions and the location of the gratings. Cw threshold currents of 10.5 mA, slope efficiencies of 0.21 W/A, and extinction ratios of 18 dB at a modulator reverse bias of 1.0 V are achieved for these devices.

The design and operation of multiple-quantum well (MQW) wavelength tunable distributed Bragg reflector (DBR) lasers with nonabsorbing gratings and monolithically integrated external cavity electroabsorption modulators fabricated by selective-area metal-organic chemical vapor deposition (MOCVD) are presented. Uncoated devices exhibit cw threshold currents as low as 10.5 mA with slope efficiencies of 0.21 W/A from the laser facet and 0.06 W/A from the modulator facet. After the application of facet coatings, slope efficiencies from the modulator facet increase to 0.14 W/A. Wavelength tuning of 7 nm is obtained by injection current heating of the DBR section. These devices exhibit extinction ratios of 18 dB from the modulator facet at a low modulator bias of 1 V, when measured with a broad-area detector. When coupled to a singlemode fiber, these devices exhibit high extinction ratios of 40 dB at a modulator bias of 1.25 V. Photo-generated current versus optical power plots indicate that the extinction ratios are not limited by carrier build- up in the modulator quantum wells.

We discuss the fabrication and optical characterization of strained-layer InGaAs-GaAs nanometer scale wire arrays grown by selective-area MOCVD on silicon dioxide patterned substrates. The wire patterns studied were obtained by high resolution electron beam lithography on PMMA using a silicon dioxide lift-off process. The dependence of the growth structure on the wire orientation is presented. Wire arrays aligned parallel to the [011] crystal direction are found to be the extremely useful for the growth of narrow quantum wire structures. Due to the faceted nature of the growth, a large non-linear enhancement of growth inside the wire region is observed. In addition, the results of gas phase diffusion growth simulations on the expected inhomogeneity of the fabricated quantum wires are presented. The degree of inhomogeneity of fabricated quantum wire arrays was studied by spatially resolved photoluminescence. Our results show that a suitable patterning technique, coupled with proper growth conditions, could allow control of the selective growth profile across the wire array. Finally, the growth of strained wires with a lateral dimension of less than 50 nm is displayed along with optical characterization of the quantum wires by low temperature photoluminescence.

Polarization insensitive 1.5 micrometer QW optical amplifiers, modulators, and detectors were fabricated using a novel, simple, post-growth, integratable technique. The process utilizes ion-implantation-induced, spatially selective, quantum well (QW) shape modification. A simple model shows that if the interdiffusion rate of the anions is larger than that of the cations, the blue shift in the ground state heavy hole transition energy after implantation and annealing is greater than the light hole state blue shift, merging the two bands and thus eliminating the difference between the TE (transverse electric) and TM (transverse magnetic) waveguide propagation modes. Current- voltage measurements indicate that junction characteristics are well maintained after processing. This simple technique for fabricating discrete polarization insensitive optoelectronic devices is readily extended to the monolithic integration of such devices along with other passive and active optoelectronic devices and provides a pathway to practical photonic integrated circuits.

DFB laser integrated with external MZ modulator is an excellent candidate for long haul fiber optical communication systems with modulation speed beyond 10 GBits/s. An understanding of the behavior of the adiabatic chirp is required for device optimization. Adiabatic frequency chirps under various conditions are calculated based on a model in which the influence of the reflection from MZ modulator to the monolithically integrated DFB laser is considered. For a given modulator structure, the results show that adiabatic frequency chirp can be reduced by a proper choice of DFB laser parameters.

Theoretical analysis and design principles for a polarization-independent optical wavelength filter based on phase-shifted gratings are developed and presented. Side- lobe suppression is demonstrated by chirping the ratio of the length of phase-shifted section to that of grating section. A design guideline is explained for the polarization-independent filter with side-lobe suppression. It is shown that identical spectral responses for the two polarizations can be achieved despite the polarization- dependent coupling strengths. Tunability of such devices is studied, and the polarization-dependent modal loss is discussed at last.

The integration of MSM photodetector arrays with polyimide ridge waveguides is demonstrated. MSM detectors consisting of two Schottky interdigitated electrodes were fabricated singly and in arrays of two or four, on semi-insulating GaAs substrates. Following deposition and patterning of an SiO₂ buffer layer, polyimide ridge waveguides were fabricated on top by spin coating and photolithography. The guides were multimode, with widths from 10 to 50 micrometer, allowing for ease of coupling from an optical fiber. Light from the waveguides was coupled through gaps in the SiO₂ buffer layer into the photodetectors. Transparent indium tin oxide (ITO) Schottky electrodes were employed to maximize absorption of light in the detector region. The end-to-end responsivities of the integrated MSM devices were typically 0.1 to 0.16 A/W. Bandwidths were 1 to 1.7 GHz; however these values could be increased substantially by optimization of the etch conditions used in the detector fabrication. Losses due to butt coupling to the multimode waveguides were around 1.5 dB. Division of the input signal between sets of two and four detectors has been demonstrated using a series of optical taps fabricated in an overlying polyimide ridge waveguide. Results indicate that polyimide waveguides could be a practical means of monolithically integrating optical functions such as signal routing and power division on complex optoelectronic integrated circuits (OEICs).

WDM networks offer potentially increased bandwidth and functionality when compared to single wavelength ones, with applications ranging from straightforward capacity enhancement through to switched virtual dark fiber carrying disparate traffic to the customers' premises. However, to be implemented in the ground, these optical networks have to offer cost effective solutions to the users. This means that WDM equipment must be built small, rugged with low power consumption and high functional density. The NAM (network access module) developed by the ONTC, makes extensive use of array electronics, optoelectronics and optoelectronic integrated circuits (OEICs) and illustrates such system integration. But high level integration and high functional density bring new challenges in packaging together with the difficulty of managing crosstalk. In this paper we describe the network access module, detail some of the key packaging technologies and discuss alternative integration approaches which are explored as part of another collaboration project called Rozinante. To conclude, some applications using the NAM are proposed which show the key role of this network element in building reconfigurable optical networks.

Optical delay-line networks are investigated for electrical signal filtering applications. Results concerning the microwave frequency response of some basic structures such as unbalanced Mach-Zehnder interferometers and ring resonators based on fiber-optic components are briefly summarized. We report the microwave performance of some integrated unbalanced Mach-Zehnder interferometers fabricated in glass by a silver ion-exchanged process. These devices can be used as microwave notch filter in the 1.5 GHz - 15 GHz range. At the resonance frequencies, a rejection depth of over 30 dBe was obtained. Furthermore, we focus on the interference effects of the light carrier which can occur, even in the incoherent regime, when basic structures are cascaded in order to perform complex filtering responses. These include the role of optical phase shift and polarization dependence. A new transfer matrix analysis is introduced for modeling the microwave response of cascaded devices showing good agreement with experiments.

Throughout the paper, novel all-optical planar 1-stage k multiplied by k-switches and compact minimum-stage k multiplied by k-switches in double-layer and multi-layer technique, are presented and analyzed. In the first case, the number of k(k - 1)/2 switches of size 2 multiplied by 2 (equivalent minimum of the Spanke-Benes network) are arranged in parallel instead of the number of k (equivalent maximum) cascaded 2 multiplied by 2-switches of the Spanke- Benes network. In the second case, the number of 2 multiplied by 2-switches depends on the geometry of the 'pipes' of the switches formed by the layers and waveguides [for a square it is 3k/2(k/2 - 1) for rearrangeable nonblocking and 3(k - 1)k/2(k/2 - 1) for circuit switching networks]. The number of stages (NS) (horizontal cascaded) of the proposed compact switches for the nonblocking interconnection is NS equals n - 1 if the waveguides form an n-gon (n greater than or equal to 3) for any size of the k multiplied by k-switch. In this way, the attenuation of optical signals passing through a photonic network may be minimized. In particular, for any size of a k multiplied by k-switch, dependent on the n-gon, the minimum NS is n-1 equals 2 (triangle) or n - 1 equals 3 (square) etc. Thus the proposed switch concept is of complexity O(1), i.e. the NS is independent of the number of inputs/outputs. Additionally, the proposed switches are capable to operate in the circuit switching mode if and only if (iff) the parallelism increases by the factor k-1.