Significant advancements in high-speed lithium niobate modulators have ushered in the age of 10 Gb/s DWDM optical transmission systems. In this paper, the performance characteristics of these integrated optical devices will be discussed with special emphasis on the wavelength dependence of key parameters. The broadband response of lithium niobate makes it well suited for high-speed external modulation, while the well-established properties of this modulator insure that performance is both predictable and reproducible. Also included in this paper will be an assessment of the future use of lithium niobate devices.

A rigorous numerical model has been developed to study Ti:LiNbO₃ modulators. In this study two types of devices are examined: directional coupler-based and Mach-Zehnder interferometric-based modulators. The benefits of using etched ridge waveguides for the MZI modulator are also shown: namely a significant reduction in V_(pi )L and smaller, more compact bends for the input and output waveguides to the MZI modulator without incurring as high loss when compared with an unetched waveguide modulator. In this study it was then shown that the device characteristics of these modulators can be optimized by variation of some of the fabrication parameters.

This paper describes the recent progress and status of 40 GHz high-speed LiNbO3 optical modulators, and newly designed two types - a backslot type and ridge type - LiNbO3 optical modulators with high-speed and low-switching voltage. The backslot is formed on the backside of LiNbO3 substrate by using micro-machining laser etching. The backslot type modulator is designed, fabricated and characterized. The ridge type modulator with an overhanged upper electrode is also calculated. The properties of the designed two types modulators, effective refractive index for the modulation wave nm, modulation bandwidth fm, overlap integral and switching voltage V (pi) are calculated. The optimum properties of the backslot type modulator, fm equals 73GHz and the switching voltage V (pi) equals 2.8V is calculated, and fm equals 130 GHz and V (pi) equals 1.9 V is achieved for the ridge type modulator at a wavelength 1.5 micrometers . The fabricated backslot type modulator achieved the optical 3dB bandwidth fm >= 28 Ghz and the driving voltage V (pi) equals 2.8 V.

SSB digital modulation is considered to be an important candidate to break through into 0.8 pbs/Hz for long hauls. Unwanted nonlinearity effects during transmission will be reduced because the carrier component is also eliminated by the modulation. The introduction of SSB scheme is a starting point to go into the synchronous detection instead of present IM-DD systems. Details of our SSB modulator is also be shown and discussed will be possible system applications of SSB modulation.

Electro-absorption modulated sources are likely to be key components in the evolution of optical communication line rates from 10Gb/s to 40 Gb/s. Compared with the LiNbO₃ alternative EA modulators are more compact, less expensive, compatible with monolithic integration, and offer lower drive voltages. However, fabrication complexity and open questions concerning the fidelity with which they transmit information make the exact role of 40Gb/s EA modulators in advanced communication systems somewhat unclear. In this talk we will describe the design, fabrication and transmission performance of 40Gb/s EA modulators configured for both NRZ and RZ operation. For NRZ transmission the device structure consists of a short MQW modulator with spot-size converters on the input and output ends. Tandem EA modulators for pulse carver and data encoder functions were monolithically integrated along with a semiconductor optical amplifier and input/output spot-size converters to explore RZ transmission. Both single and tandem modulator design are realized using semi-insulting InP current confined buried heterostructure technology. Modulation bandwidth of better than 50 GHz is demonstrated along with a fiber-to-fiber insertion loss of less than 6 dB for the single modulator design. The carver/encoder configuration with onboard SOA yields better than 0 dB insertion loss. Transmission impairments were studies using both designs.

While tunable lasers have been a focus of research and development efforts for over 10 years, they have only recently gained market acceptance in optical transport and networking. Tunable lasers offer many compelling advantages over fixed wavelength solutions in optical networks in that they reduce inventories, allow dynamic wavelength provisioning, and simplify network control software. More interesting, is that tunable lasers have been featured in optical network development efforts in every segment: access/enterprise, metropolitan, and long haul networks leading to a variety of desired specifications and approaches. In fact, the term 'tunable laser' has come to describe an increasingly broad range of technologies from monolithic semiconductor lasers, to MEMS (Micro-Electro-Mechanical Systems) based lasers and fiber lasers. This presentation will focus on monolithic, widely-tunable lasers which are promising candidates to satisfy the needs of all the market segments mentioned.

As carriers and service providers continue their quest for profitable network solutions, they have shifted their focus from raw bandwidth to rapid provisioning, delivery and management of revenue generating services. Inherently transparent to data rate the transmission wavelength and data format, MEMS add scalability, reliability, low power and compact size providing flexible solutions to the management and/or fiber channels in long haul, metro, and access networks. MEMS based photonic switches have gone from the lab to commercial availability and are now currently in carrier trials and volume production. 2D MEMS switches offer low up-front deployment costs while remaining scalable to large arrays. They allow for transparent, native protocol transmission. 2D switches enable rapid service turn-up and management for many existing and emerging revenue rich services such as storage connectivity, optical Ethernet, wavelength leasing and optical VPN. As the network services evolve, the larger 3D MEMS switches, which provide greater scalability and flexibility, will become economically viable to serve the ever-increasing needs.

We present a brief overview of a promising switching technology: Silica-on-Silicon thermo-optic planar lightwave circuit integrated circuits (PLCs). This 2-D solid-state optical device is capable of non-blocking switching operations and several additional important built-in functionalities. Both enable single-to-single channel switching, and single-to-multiple channel multicasting/broadcasting. In addition, it has the capability of channel weighting and variable output attenuationlpower control, for instance, to equalize signal levels and compensate for unbalanced different optical input powers, or to equalize unbalanced EDFA gain curve. We mention the market segments appropriate for the switch size and technology, followed by several application examples: (1) Core networks use cross-connect systems to establish connections among nodes as well as among network segments; (2) Switching to protect and restore network service.

We begin by reviewing the likely evolution of long-haul optical networks, and the role of optical cross connects (OXCs) in this evolution. We examine technologies for the switch fabrics in these OXCS, and focus on a guided-wave optical crossbar matrix based on total internal reflection. We argue that the reliability, modularity and scalability of crossbar matrix switches make them attractive for many applications.

In this work, we demonstrated several different types of liquid-crystal WDM (wavelength-division-multiplexing) signal processors including broadband optical switches, dynamically-variable optical attenuators and optical harmonic equalizers.

We have developed a 2 by 2 digital optical switch (DOS) composed of four 1 by 2 DOS elements and a thermo-optic wavelength tunable filter using silicone resin waveguides. The 2 by 2 DOS operated with a switching power of 380 mW, a very low crosstalk of less than -60 dB, and a low insertion loss of less than 2.5 dB at 1.55 micrometers . The tunable filter operated with a maximum power of 2.5 W, a tuning range of > 30 nm, a crosstalk of < -30 dB, and response time of < 60 ms. We also applied our polymer tunable filter to a WDM/SCM LAN system to confirm its practicality.

This report reviews optical switching technology using electro optic material. Firstly, guided wave optical switch using single routing stage and then deices implemented by connecting multiple ranks of switching elements are described. The number of modes used in operating and the device structure can classify switching elements. Representative switching network architectures constructed by connecting switching elements are shown. Some device demonstrations using lithium niobate and other materials such as compound semiconductor are introduced.

Current performances and future prospects of functional planar lightwave circuits (PLCs) will be described. Channel numbers of AWGs have been dramatically increased up to 400ch in single wafer. In the multi-chip configuration, 1010 ch has been achieved with 10 GHz channel spacing. Although throughput of the transport can be increased to tera-bit level by the WDM technology, signal processing in the electrical layer becomes a serious problem due to its sped limit. Optical functional devices, therefore, are important for solving these issues. Various kinds of optical signal processing devices have been developed; they are, dispersion slope equalizers, PMD equalizers, temporal pulse shapers and optical label recognition circuits. Optical address signal recognition will be a key technology for future label- switched networks, where each router has to provide an ultrahigh throughput exceeding the electronic speed limits. We proposed a novel optical circuit for recognizing an optical pulse pattern, which is based on an optical digital- to-analog converter fabricated on a PLC. The circuit converts the optical pulse pattern to analog optical amplitude, and enables us to recognize the address.

We report some of our recent results on all-optical wavelength conversion including result on broadband orthogonal pumped four-wave-mixing in semiconductor optical amplifiers and dual-wavelength injection-locking of a Fabry- Perot laser diode. The high speed performance of wavelength conversion buy dual-wavelength injection locking of a Fabry- Perot laser was investigated experimentally and we present for the first time results on wavelength conversion at 10Gbit/s using dual wavelength injection-locking of a Fabry- Perot laser diode. The experimental results for all-optical wavelength conversion using broadband orthogonal pumping in a fiber ring containing a semiconductor optical amplifier is described. We also describe the latest result on a 40Gbit/s polarization independent all-optical wavelength converter based on polarization diversity and four-wave-mixing in a single semiconductor optical amplifier.

This paper presents the recent progress on several kinds of photonic devices for WDM applications. Wavelength-tunable light sources for optical transmitters, wavelength selectors for optical cross-connect, and wavelength converters fro WDM routing will be discussed.

Semiconductor optical amplifier (SOA)-based functional devices are of interest for optical signal processing due3 to their many attractive features, which include high-speed operation, potentially low polarization dependency, large optical bandwidth and low loss. Furthermore, they may be integrateable and have a moderate to low power consumption. The fact that SOA-based devices can be used for a multitude of purposes and either have the above-mentioned characteristics inherently or can be fabricated to accommodate them, makes them prime candidates for the implementation of a variety of functionalities. Among others, SOAs are attractive as fast optical gates, where one of their most significant features is a high on-off ratio, which is critical for cross-talk suppression. Furthermore, SOAs employed as nonlinear elements, either alone or in interferometers, have demonstrated the capability of more complex functionalities such as wavelength conversion, switching, time-division (de)multiplexing as well as 2R and 3R regeneration. Additionally, more exotic functionalities such as Boolean logic gates can be realized, making SOA- based devices attractive for use in a broader system or network perspective. In this paper, the focus will be on techniques for wavelength conversion, 2R and 3R regeneration employing cross-phase modulation in interferometers, which have demonstrated some of the most promising results. Furthermore, the implementation of all-optical logic using SOA-based interferometers will be discussed, and some of the novel employments of these functionalities will be presented.

As the demand for optical fiber communications bandwidth grows, the implementation of signal processing functions using all-optical techniques becomes increasingly attractive. In recent years, a number of methods have been used to perform functions such as wavelength conversion for WDM systems, gated mixing for TDM multiplexing and demultiplexing, spectral inversion for dispersion compensation, and all-optical switching. Three-wave mixing in c(2) media is an attractive approach, presenting a combination of low pump power, wide bandwidth, and negligible degradation of signal to noise ratio. In this paper, we describe optical frequency mixers implemented using annealed proton exchanged waveguides in periodically poled lithium niobate. These devices have been used in a variety of system experiments. We present several WDM demonstrations, including wavelength conversion, dispersion compensation by mid-span spectral inversion, and compensation of Kerr nonlinearities. We also discuss TDM demonstrations such as efficient all-optical gating and multiplexing/demultiplexing of high bit-rate data streams.

The state of the art InP/InGaAs avalanche photodiodes (APDs) and the R and D status of the novel APDs are reviewed. Highly sensitive and reliable 10 Gb/s photo receivers consisted of the conventional InP/InGaAs and InGaP/GaAs heterojunction bipolar transistor ICs are very practical and are now in volume production. The novel APDs reveal high performances such as high-speed and low-noise characteristics, indicating great potential for application to the ultra high-speed transmissions toward 40 Gb/s regime. Very recent study on the strain compensated multiple quantum well APDs aimed at L-band DWDM systems applications over 1.6micrometers wavelength is also introduced.

The absorption layer of a strain-compensated InGaAs multiple quantum well (MQW) was shown to improve the responsivity at the L-band wavelength of a photodiode used in optical fiber communications. The MQW was examined to clarify the range of structural parameters, which are strain and thickness, with which a smooth surface morphology can be obtained. A photodiode with an absorption region of a strain-compensated MQW exhibiting a smooth surface morphology was fabricated, and it was proven to have a high crystalline quality with a low dark current. The MQW absorption region enabled the responsivity of the photodiode to exceed that of a lattice- matched InGaAs photodiode. Carrier transport vertical to the MQW layer was shown to be scarcely affected by the hole trapping caused by band discontinuity.

As the demand for increased bandwidth continues to grow, the implementation of 10 Gbit/s systems is underway and telecommunications companies are now beginning to see a need for 40 Gbit/s per channel. One of the key enablers for building such 40 Gbit/s systems is the availability of high bandwidth receivers. At 40 Gbit/s the fundamental tradeoff between the transit time limited bandwidth and the internal quantum efficiency for conventional surface illuminated PIN detectors makes them unsuitable. It is therefore necessary to employ a waveguide photodetector design. Many 40 Gbit/s systems employ return-to-zero transmission formats and polarization interleaving. This added complexity places additional demands on both the bandwidth and the polarization dependence of these detectors. The high demands placed on the detector design make an optoelectronic integrated circuit approach for the receiver very difficult to realize. In this paper, we will discuss a 40 Gbit/s receiver with a responsivity greater than 0.7 A/W and polarization dependence of less than 0.1 dB over the wavelength range from 1500 nm to 1620 nm. This receiver has a 3dB bandwidth of 40 Ghz, a conversion gain as high as 120 V/W, < 20 pA/(root) Hz input referred noise, < -7 dBm sensitivity, low deviation from linear phase, and a linear output swing greater than 0.8 V p-p.

We describe a new photonic-crystal structure with a complete three-dimensional photonic band gap (PBG) and its potential application to integrated optics. The structure not only has a large band gap and is amenable to layer-by-layer litho-fabrication, but also introduces the feature of high-symmetry planar layers resembling 2D photonic crystals. This feature enables integrated optical devices to be constructed by modification of only a single layer, and supports waveguide and resonant-cavity modes that strongly resemble the corresponding modes in the simpler and well-understood 2D systems. In contrast to previous attempts to realize 2D crystals in 3D via finite-height 'slabs', however, the complete PBG of the new system eliminates the possibility of radiation losses. Thus, it provides a robust infrastructure within which to embed complex optical networks, combining elements such as compact filters, channel-drops and waveguide bends/junctions that have previously been proposed in 2D photonic crystals.

An AlGaAs-based near IR 2D photonic crystal with an air- bridge structure featuring defect waveguides has been developed. For the sample without defect waveguides, measurement of the optical transmission characteristics in the wavelength ranging form 850 nm to 1100 nm showed a deep bandgap with 30 dB to 35 dB attenuation and transmittance of nearly 100 percent. Optical propagation properties of defect waveguides were obtained by two methods: measurement of transmission spectra at wavelengths ranging from 850 to 1100 nm, and with plan-view observations of the optical beam trace along the waveguide measured with an IR-vidicon camera. 3D FDTD simulations for the band structure and transmission spectra in the air-bridge slab with and without defect waveguides resulted in the appearance of four defect propagation modes specific to the defect waveguide, between two slab modes for the defect-free photonic crystal slab. These defect modes were experimentally identified in the measured transmission spectra. Low group velocity near the band edge was also confirmed through 200-fs pulsed laser- beam propagation measurement.

Using an asymptotic matrix formalism, we analyze the guided modes of Bragg fibers and the dielectric coaxial fibers. In the asymptotic limit, the Bloch theorem can be applied to describe the optical field within the cladding layers, while the core region field is described by the exact solutions of Maxwell equations. From the asymptotic analysis, we derive an approximate expression for the radiation loss of Bragg fibers and dielectric coaxial fibers and give the number of Bragg pairs required to achieve 0.2dB/km radiation loss. The dispersions of the guided modes of Bragg fibers and dielectric coaxial fibers are calculated using both the asymptotic approach and the finite difference time domain method. The results obtained from these two approaches are shown to have excellent agreement. We use asymptotic analysis to calculate the dispersion parameter D of the guided dielectric coaxial fiber modes, which is found to be much larger than that of the conventional telecom fibers.

In 1996 we reported the first example of a photonic crystal fibre (PCF), an entirely new class of optical fiber. Also known as holey or microstructure fibers, they incorporate air holes that run along the length of the fibers cladding. The fiber is made from a stack of close-packed silica tubes and rods that is drawn into fiber using a conventional fiber drawing tower. We have demonstrated a wide variety of PCF designs and developed the conceptual tools needed to understand their properties and guide their design. These fibers can have highly unusual properties, and look set to rewrite the fibre-optics rulebook and revolutionize the future of optical telecommunication.

The technology ofBragg grating formation has come a long way since the initial demonstrations back in 1978 [1]. Techniques capable of forming nearly any refractive index modulation and imprint nearly any grating pitch at a given position along the length ofthe grating are now in widespread use [2]. This evolution in manufacturing techniques has lead to an impressive collection of a variety of Bragg grating based components and devices, [3], and the robustness of many of these devices in laboratory experiments have pushed their employment in real system environments. Convincing performance, in the relative short lifetime of the technology, led in 1995 to the commercialisation of a series Bragg grating products. Among these were the chirped fibre Bragg grating. It was originally proposed by Ouellette [4] as a device capable of compensation of chromatic dispersion by incorporating reflective differential delay characteristics for different wavelengths along the grating. The pulse broadening suffered when transmitting signals through installed telecommunications fibre therefore can be compensated by delaying the different spectral components of the pulse such that it is reformed to its original shape upon reflection from the grating. Although chromatic dispersion does not impose much limitation to signal recovery when operating systems at bit-rates less than 10 Gbit/s and such data-rates were far from being realised when the chirped fibre grating was proposed, it was foreseen to play an important role in the then future high bit-rate transmission systems. As data-rate transmission at 10 Gbit/s now indeed are beginning to be a well-established technology, and even systems employing wavelength division multiplexed (WDM) technology at the same bit-rate are working solidly, there is a clear need for robust dispersion management in these systems. As a chirped grating exhibits many attractive features such as compactness, low insertion loss, and very importantly, low sensitivity to non-linearities together with low or no polarisation dependancy, it is a leading contestant compared with other proposed dispersion management techniques. These include the use of dispersion compensating fibre (DCF) [5]. The main advantage of DCF is the broad bandwidth over which dispersion compensation can be performed, but high sensitivity to non-linear effects and the missing ability to "shape" the dispersion profile together with the lack of tunability, are major drawbacks to this approach. The previous lack of bandwidth of chirped gratings has been the only limiting factor for them in being the most interesting, simple and robust approach to chromatic dispersion compensation. The design and realisation of high quality truly broadband gratings therefore has taken highest priority among grating manufactures. We will in this paper from a design point of view discuss how chirped gratings with large bandwidths and high quality dispersion profiles can be achieved. Furthermore, examples of gratings with narrowband spectral responses will be given -gratingsdesigned using a new powerful inverse-scattering technique. One approach to the fabrication of very long and complex chirped gratings will also by aired supported by a number of experimental grating examples using this technique. A discussion of how imperfections imposed by either the manufacturing process of -ora non-perfect waveguide environment to the gratings, can be minimised by choosing the right host fibre parameters, will also be given. Finally techniques for tuning ofthe dispersion ofBragg gratings will be discussed together with recent advances in the experimental demonstrations of these.

The amount of uncompensated dispersion that is tolerated in optical communications links decreases sharply with increase in channel line-rates. Since this holds for each channel individually, multi-wavelength WDM transmission at high bit-rates requires simultaneous dispersion compensation over the entire spectral range of interest. This entails compensating for the dispersion slope as well as the dispersion of the transmission fibers used in the link. The class of non-zero dispersion-shifted fibers (NZDSF), which are widely deployed today, exhibit high relative dispersion slopes. Thus, dispersion management of 40 Gb/sec links comprised of NZDSF poses a significant challenge. This talk will introduce a novel dispersion compensation scheme that utilizes propagation in a higher-order mode of a few-moded fiber. The primary advantage of higher order modes is that they exhibit larger dispersions as well as dispersion slopes. This facilitates broadband compensation for any transmission fiber among the class of NZDSF. In addition, these modes have larger effective areas, and hence reduce non-linear distortions in the transmitted signal. Consequently, this compensation technique provides the additional benefit of maintaining higher signal powers throughout the span, which in turn leads to longer transmission distances. Recent breakthroughs that allow for low loss, broadband mode-conversion have made this technology viable for practical applications. The physics and characteristics of this technology will be described, and its attendant systems advancements will be compared with that of alternate schemes.

A system for generating polarization mode dispersion has been developed to reproduce in the laboratory, a close approximation to the polarization mode dispersion (PMD) of fiber links. A multi-stage system of rotating wave plates and combinations of birefringent elements generates a mixture of first order, second order and higher order PMD suitable for testing OC-192 and OC-768 hardware. The PMD Emulator can be operated in either static or time-varying modes. In the static mode first order PMD is generated with the Differential Group Delay continuously adjustable from 0 to 160 ps. In the dynamic mode time-varying PMD containing first and higher orders is generated with an approximately Maxwellian DGD distribution. The PMD emulator generates PMD variations at adjustable frequencies up to 400 Hz.

PMD is one of the critical hurdles in deploying high capacity fiber optical networks (OC-192 per channel and above). Due to its stochastic and time-varying nature, PMD compensation requires adaptive schemes, using feedback or feed-forward techniques. In this paper we will discuss practical issues and limitations of PMD compensation, as well as recent developments on architecture and feedback schemes.

We report a new configuration of tunable optical filter for DWDM applications. In this design, first-order diffracted signal light by a grating is directed to a lens and focused on to a transmission-type liquid crystal spatial light modulator (LC-SLM). Wavelength channels are selected by opening the appropriate pixels of the LC-SLM for transmission. The device is demonstrated by using a broad-band AR-coated laser diode as the signal light. For conceptual demonstration, we use optical fibers with a core diameter of 50 microns instead of the LC-SLM. A total of 37 channels, spanning 33.6 nm were successfully selected with channel spacing of 0.94 nm and bandwidth of 0.4 nm. With 100 micron-wide pixels separated by 5 microns in the LC-SLM and a 1800 lines/mm grating, we show selection of five channels with channel spacing of 0. 78 nm. The bandwidth is 0.43 nm. The channel isolation is better than 20 dB. At a wavelength of 1.5 micron, channel spacing as small as 12.4 GHz can be realized. This device is also expected to be useful for other DWDM applications, e.g., switching and routing.

The design and manufacture of advanced telecommunication devices based on Fiber Bragg gratings (FBG) is presented. With the maturing of optical communication comes the need for advanced optical filtering capabilities that permit the direct manipulation and processing of optical data streams in both the wavelength and time domains. By creating very complex grating amplitude and phase profiles inside fibers it is possible to approach an arbitrary transfer function capability. We demonstrate that with these filters it is possible, for example, to improve greatly the filtering characteristics vis-a-vis WDM or chromatic dispersion compensation applications. More importantly, more advanced filtering can be achieved in which temporal codes can be imprinted directly inside the optical bits and subsequently recognized by matched optical filtering. This optical code-division multiplexing (CDM) capability presents interesting new opportunities for advanced optical processing allowing for increased channel count, novel optical routing abilities, or more flexible bandwidth distribution. Moreover, the filtering can be made compatible with WDM and TDM systems, and hence takes a large step towards enabling the introduction of more complex modulation formats in optical communication systems. An important consequence of the arbitrary filtering capability is that multiple filtering functionalities can be combined inside a single FBG. In this context, we present a filter that combines wavelength selectivity with code discrimination and dispersion compensation.

Bragg grating-based optical systems are important for both telecommunications and sensor applications. Work to date on the simulation of such systems has concentrated upon using approximate methods such as the coupled mode theory (CMT). In this work, a combination of three numerical methods has ben used, all of which are rigorous and at the same time computationally very efficient. The new approach presented here incorporates the finite element, the least square boundary residual and the transfer matrix methods. The simulated results obtained show that the CMT is in general adequate for the characterization of Bragg grating devices in fiber, since the perturbed refractive index change is small. However, for Bragg grating devices in semiconductors, the CMT is shown to generate less accurate result. Simulated results obtained for various types of grating devices, such as uniform, chirped, apodized, phase-shifted, super- structures and sampled grating devices are presented.

The basic design methodology and criteria required for packaging of optical components are reviewed, and the state-of-art of different types of the packaging technologies of laser modules and receiver modules are addressed. They are Mini-DIL DFB-LD module, wavelength-locker integrated laser module, 10 Gbps receiver module with integrated APD and preamplifier IC, and non-hermetic SMT module. Considerations on future packaging technologies and integrated optical components are also discussed.

A rigorous numerical approach, based on the full vectorial finite element method, is used to design various types of monolithically integrated spot-size converters for efficient coupling to a standard optical fiber. Spot-size converters with a tapered core, using uniform directional couplers, and MMI structures are discussed and results of performance reported.

Polymer waveguides provide cost effective interconnect solutions for high-volume applications required by the rapid growth in VCSEL array sizes and product demand. Multimode polymer waveguides 34-channels wide have been stacked in arrays 12-layers high with center-to-center waveguide spacing of 125 +/- 2 microns between layers and 90 +/- 2 microns within a layer. No measurable crosstalk between channels has been observed even when separation between multimode waveguides was reduced to a 4-micron gap. Flexible polymers provide an out-of-plane bend radius of less than 5 mm that simplifies VCSEL packaging requirements and volume. Transitioning the waveguide pitch within and between the polymer layers from 125 to 250 microns enables interface of high-density VCSEL arrays to standard fiber ribbons. Passive fiber pigtailing to 62.5/125 fibers was achieved with < 0.5 dB loss. Pigtailing can be avoided entirely by direct connectorization of the polymer waveguide arrays with industry standard MT connectors. Optical CrossLink's Guidelink polymer waveguide devices are made form sheets several hundred feet long. Waveguide's are formed using contact photolithography that requires no costly spin- coating, wet chemistry, embossing, modeling or etching techniques required by other planar waveguide fabrication processes. All required processes are suitable for automation with high-yield while at the same time drastically reducing the infrastructure required to produce devices. Currently, the equivalent of 100 six-inch wafers of planar waveguides can be produced in less than 2 days without automated machinery.

This paper reviews III-V semiconductor dry etching technologies established in the past decade for miniaturizing and integrating photonic devices/components and nano-fabrication under development for creating novel photonic structures such as photonic crystal and quantum dots. After briefing the technology requirements for DWDM/OTDM-based Terabit optical communication era in 2005- 2010, advancement of the GaAs- and InP-based smooth and high-aspect-ration dry etching with micrometers -size is reviewed with some applications to dry-etched laser diodes and waveguide devices. Secondly, EB nano-lithography and dry etching technologies for 10- to 100-nm-size structures are reviewed for demonstrating photonic crystal. Challenging application to extremely miniaturized waveguide-based planar light wave circuits is included. Lastly, nano-probe assisted processing of arrayed quantum dots as a 10-nm-size structure is discussed. Achievement of suppressed size fluctuation using this technology will prov8de us with a possibility of large optical non-linearity promising for all-optical switching devices in the OTDM optical communication network system.

We have developed new thin optical multi-fiber connector, which applies MT connector and is appropriated for high dense arrange. This thin-type connector consists of housing that have two latch claws for connecting to a thin-type receptacle and spring that applies pressing force to fiber endfaces. It is a feature that connection and disconnection are operated by hand. It is confirmed that the optical characteristics are favorable to be equivalent to standard MT connectors engaged by MT clip.

It is shown that fiber surface performance degradation due to scratches and pits can be analyzed using wave scattering by imperfect surfaces. The return loss of a fiber endface/interface is obtained in terms of a parameter which gives the relative increase in scattering due to a defect over a defectless fiber endface. This parameter can be expressed as the ratio of the bidirectional scatter distribution functions with and without the defect. The predicted return losses as a function of the ratios mentioned above are presented.

Optical amplifiers are demanded to improve their performances in order to support the forthcoming Internet oriented WDM networks. This paper reviews recent topics on next generation optical fiber amplifier technologies developed at author's research group.

Present efforts towards construction of large capacity long-haul links are focused on either increase of channel data rate or a very dense channel plans. The latter is of particular interest since it enables spectrally efficient WDM links simply by multiplying the channel count of existing transmission lines. Unfortunately, very dense channel spacing necessarily leads to increased four-wave (FWM) mixing penalty. FWM penalty can be decreased in straightforward manner by lowering signal launch power. Indeed, high spectral efficiency optical link operating with 25GHz spaced signals has been recently demonstrated using hybrid Raman/EDFA amplification architecture: FWM process is minimized by operating in quasilinear transmission regime1. A different approach that does not require deployment of Raman amplifiers relays on combining two closely spaced counterpropagating interleaved channel grids.2 We will describe fundamental design issues encountered in such dense bidirectional optical links. While bidirectional link potentially doubles deployed capacity over single strand of fiber, its realization is made more difficult by host of qualitatively new problems, commonly not seen in conventional unidirectional lines. We describe latest efforts in constructing very dense bidirectional long-haul links over standard and non-zero dispersion shifted fiber.

High power pump lasers in wavelength range from 1400 to 1520 nm, namely 14XX nm lasers, are heavily demanded for Raman amplification as well as existing Erbium doped fiber (EDF) amplification in recent Dense Wavelength Division Multiplexing (DWDM) optical communication networks.We achieved the record highest optical power output 14XX nm pump laser of 400 mW at the single mode fiber end as a production level. Rollover fiber coupled power was over 500 mW for this device. The pump laser module with newly designed package operated up to the case temperature of 75 degrees C, and laser chip temperature of 25 degrees C. This is the first demonstration at the operating case temperature as high as 75 degree C with 400 mW optical power output range of 14XX nm pump laser module. The technical challenge for higher optical power output operation is how to minimize the heat-generation from the laser chip. For this purpose, we investigated two different design parameters: the operating power consumption and the operating current of the laser.We found that the former parameter is more effective design goal than the latter one, in order to realize both high optical power output operation and long term reliability performances. We also showed that the reliability performance in 14XX nm lasers is not dependent on the facet optical power output, but dominated by the junction temperature at the active region of the lasers.

We have demonstrated the self-align electrode separation technique for directionally coupled semiconductor optical amplifiers (DC-SOA). Using oblique electron-beam evaporation, the electrodes between the coupled waveguides of the DC-SOA were separated without any lithography. We have obtained isolation resistance between the electrodes of about 130 (Omega) , and satisfactory L-I characteristics of the DC-SOA were measured. We have also shown all-optical flip-flop operation using the electrode separated DC-SOA by numerical analysis, and the lasing threshold current control by optical injection was experimentally demonstrated, thus confirming the possibility of all-optical flip-flop operation.

In recent years there has been rapid progress in the transmission capacity of lightwave systems. A review of key technological advances, which have contributed to the development of terabit capacity lightwave systems is presented in this paper. In particular, author's contribution in the ear of high capacity transmission using wideband EDFAs and Raman amplifiers is highlighted.

In order to meet ever-increasing demand for large transmission capacity, optical fiber technologies are evolving toward ultimate performance. New generation non- zero dispersion-shifted fibers, whose dispersion characteristics are optimized for ultra-large transmission capacity terrestrial networks, have recently been introduced. In the field of submarine cable networks, dispersion-flattened and dispersion-managed hybrid transmission lines are emerging rapidly.

An add/drop multiplexer (ADM) is recognized as one of the basic building blocks to extend DWDM networks from a static point-to-point system into a next generation, dynamic, re-configurable, programmable optical network. The objective of this paper is twofold: (1) provide an extensive overview tutorial of the numerous existing implementations of ADM, (2) categorize these different ADM implementations, and (3) assess their respective limitations and impacts on an evolving optical network. Toward this goal, a clear distinction between an OADM and a WADM node is made in terms of add/drop port configuration using a functional black box approach, and the types of component and device technologies that support these structures. Our second objective is to use this classification scheme to project, what we believe, is the functional form of the next generation ADM module. This is accomplished by taking into account major trends and developments in the optical networking arena. Lastly, some technical perspectives and directions toward the form of the next generation ADM are presented.

The need for flexible optical network architectures will increase as network operators attempt to support dynamic bandwidth demands. A very flexible broadband wavelength multiplexer with unique characteristics has been developed for use in flexible optical networks. This technology innovation allows flexible multiplexing of single channel traffic and multi-wavelength traffic onto one fiber. Multi- wavelength traffic spaced 100 GHz aprat modulated at 10 Gb/s is multiplexed while introducing minimal optical impairments. A significant reduction in interchannel crosstalk-like noise generated by interactions of closely spaced channels in multi-wavelength traffic is achieved.

Optical demultiplexing is an important technique for retrieving data from high speed optical transmission systems. In this technique, the data bits are first demultiplexed to much lower data rate bit streams from which the information is retrieved using conventional electronic techniques. We have performed optical demultiplexing experiments for 80 Gb/s and 160 Gb/s optical time division multiplexed signals. These experiments include data generation, clock recovery and optical demultiplexing to lower data rate bit streams. The optical devices used for demultiplexing are, LiNbO₃ modulators, four wave mixing in semiconductor amplifiers, and semiconductor amplifier based Mach-Zehnder interferometers. The high speed performance limits of these devices would also be discussed.

The advent of digital technology in HFC networks has opened up a myriad of opportunities for MSOs. The introduction of these advanced services comes at a cost: namely, the need for increased capacity; and especially increased reusable bandwidth. In HFC networks all services are ostensibly broadcast: the prime difference between services being the footprint over which these services are broadcast. Channel lineups for broadcast video services typically cover the largest are. Advertising zones are typically second, usually on the order of a typical 20K home hub. For initial penetrations for high speed data services such as cable modems, a typical hub site will be divided into several sectors using a single 6 MHz channel. Telephony services are broadcast over the smallest area, typically a 6 MHz channel for each node. Naturally as penetration of these services increase, the broadcast area for each will also decrease.

An ever-increasing bandwidth demand in computer communication and switching necessitates an efficient interconnection technology. Some of the advanced electronic parallel computer communication architectures may not even be sufficient to meet the ultra-high speed switching requirements of today's applications. The optical signal whether communicated through waveguides or free-space can propagate at the speed of light in the medium and, hence, is independent of the number of components that receive those signals. Optical interconnections are inherently fast, secure, and parallel in nature, which in turn, facilitate high bandwidth computer communication and switching. An attractive means for exploiting optical bema non-interaction property is to use free-space propagation but with focused, i.e., photorefractive volume holographic interconnection technique. In this talk we review a number of different optical interconnection network architectures with an emphasis on photorefractive volume holograms and their possibilities and problems as applied to computer communication switching.

Optical interconnections have become the medium of choice for the long haul communication. As the speed of the modern day computers are increasing, the speed of data transfer may reach a limit. This creates a motivation to replace electronic buses with all optical ones. Several avenues of research have merged in optical interconnects for digital system. In system architecture level studies, the goal is to understand the role of the different levels of interconnection hierarchy within digital system in the overall performance equation. Another avenue of studies identify active and passive components that will be most suitable to implement optical interconnects. These components are used to design the physical organization, the communication protocol in the existing computer architecture. Another type of studies may propose modification of the existing computer architecture, which can take advantage of the high bandwidth, multi-wavelength, ultrafast optical communication and various advancement in long haul optical communication.

A system for characterizing polarization controllers and other fiber components with Mueller matrices is presented. Most polarization controllers, such as lithium niobate modulators or PLZT electro-optical modulators, exhibit a wide range of polarization behaviors at constant drive voltage including elliptical retardance, polarization dependant loss (PDL), and depolarization. Specifying the half wave voltage for such devices describes their desired characteristic, an electrically addressable retardance, but not the undesired characteristics. Devices with complex polarization behaviors require a similarly comprehensive description of their polarization effects. We present example measurements that demonstrate how the Mueller matrix as a function of voltage provides a complete description of the desired retardance and the undesired PDL. Such polarization controller Mueller matrices can be multiplied with Mueller matrices for other photonic components to quantify how component polarizations interact.

In this work we apply a simple, non-destructive method we developed for the birefringence characterization of helically wound passive fibers to monomode erbium-doped optical fibers. This method is based on the Jones matrix model developed by JN Ross for helically wound optical fibers. In the case of passive fibers Ross model is correct if the polarization evolution of light is measured with respect o an input local reference frame defined by the helix geometry; but in order to use a fixed reference frame it is necessary to consider the rotation of the plane of polarization introduced by parallel transport along the fiber. The use of Poincare's method and Mueller calculus simplifies the physical interpretation of the results. The birefringence properties of two helically wound erbium fibers are tested in the neighborhood of the amplification band showing that in this case the spectral response has a much stronger variation than in the case of passive step- index monomode fibers. Despite the topological contribution due to the parallel transport of the reference frame, with method where presented provides an easy way to measure the total linear retardation induced by the fiber curvature and the total circular retardation indued by the fiber torsion. Experimental results obtained for two commercial EDF are presented.

Repetitive triangulation (RTR) of a N-gon cell complex is applied to generate a variety of active fiber bundles which, if integrated, become 2D lightwave circuits (LWCs). By means of RTR the size of existing couplers increases and new couplers are generated according to some principles. The RTR shows different results for the geometries N >= 3 and the result also depend on the degree of RTR. For the purpose of simplifying photonic integration, the large-sized couplers arising (i) at the center of the N-gon cells and (ii) interconnecting N-gon cells to cell complexes, are aimed to be reduced by the introduction of several wavelengths. The operation of the various couplers at several wavelengths represent a serious restrictions of the permutation capabilities of the switches but the applicability to optical logic is assumed. RTR may be applied to generate scalable networks of 2D LWCs which overcome the restriction of nearest-neighbor interconnections schemes and extend them to 3D LWCs. Two different space filling problems arise dependent on the chosen RTR procedure.

Binary fiber Bragg grating arrays are capable of providing a true-time-delay capability needed for beamforming. The grating arrays consists of a series of fibers connected by circulators and switches. Each fiber contains tow or more fiber Bragg gratings on a fiber may become smaller than the grating length. Alterative architectures that use ternary switching have been proposed that show promise in reducing the constraints on grating spacing, but require carful control of grating position. Component-level simulation was carried out to compare alterative architectures and to evaluate effects of tolerances in grating placement, component imperfections, connector losses, and similar impairments whose overall consequences are difficult to predict. Fiber Bragg gratings, channel balancers, switches, and other beamfomer components were modeled. Beamforming arrays were simulated that focus tow antennas, and allow beamscanning over a +/- 70 degrees angle in thirty-two discrete steps. The baseband bandwidth was assumed to exceed 1 GHz. Binary and ternary array structures were compared. The PHOTOSS photonics system simulator provided the simulation engine. Beam patterns were calculated for a beam sweeping past a broadside target, and for a beam sweeping past targets at other angles. Results of simulation have been used to inform decisions as physical systems are realized.

The current trends within ultra high-speed optical time division multiplexed (OTDM) based communication systems dictate the increasing need for all optical buffering systems. Such systems inherently avoid the bottlenecks associated with opto-electrical (O/E) and electro-optical conversions (E/O)1. These buffers would enable the storage of data for discrete time intervals, and are necessary for many OTDM applications. Storage time limitations within passive recirculating fiber loop buffers are mainly due to the dispersive, nonlinear and loss properties of the fiber. These result in both amplitude decay and pulse spreading which may have a detrimental effect on data integrity. In this paper, we examine the propagation of both standard soliton and Gaussian-soliton shaped pulses within a recirculating fiber loop buffer. The simulation model is based on the nonlinear Schrodinger equation (NLSE) and accounts for fiber loss within the communications channel. At this stage pulse interactions are not considered and direct modulation of the launched pulses is assumed. Simulation results for bit error rate performance at different buffer loop numbers is presented.