In order to increase the transmittable data rate and to enlarge the transmission distance, the spacing between WDM channels has to be decreased while the optical transparent length must be increased both giving rise to interchannel crosstalk induced by fiber nonlinearities like cross-phase modulation (XPM). Thus, the development of modulation techniques being robust towards these effects is necessary. Recently, phase shift keying (PSK) techniques have attracted remarkable interest. For PSK-techniques, the optical power as a function of time is approximately constant (for nonreturn- to-zero (NRZ) signaling) or periodic (for return-to-zero (RZ) signaling). This is an advantageous property for the reduction of nonlinear phase modulation (PM) induced by the effect of XPM. On the other hand, since for PSKtechniques the information is carried by the phase of the optical carrier, the sensitivity to the nonlinear PM is high. In our contribution, we present an analytical model for the XPM-induced PM. With the help of this model and the visualization of XPM in the complex plane, we prove that the differential self-homodyne implementation of PSK is robust towards the nonlinear PM while PSK-techniques using a local oscillator in the receiver are extremely sensitive.

The purpose of this paper is to show the dramatic increase in data throughput and spectral efficiency by applying integrated design of bandwidth efficient space-time modulation and space-time coding to guided optical transmission system. The first section introduces the basic design of a space-time fiber-optics system. This is followed by a detailed description on the design and operation of bandwidth efficient waveforms, space-time modulation, spacetime error correction code and signal equalization. Implementation, deployment strategy and performance issues are covered last.

As system architectures are pushed to the limit to provide ever-increasing capacities, 40-Gb/s DWDM appears to be the next technology that will be commercially deployed. This paper reviews some of the key enabling technologies for high capacity 40-Gb/s transmission, including distributed Raman amplification, new optical fibers, advanced modulation formats, and forward error correction. It will also explain system design trade-offs and discuss several recent large-scale system transmission experiments for multiple terabit ultra-long haul terrestrial networks.

In this paper, a nonlinear, optically amplified intensity modulated direct detection (IM/DD) system model with forward error correction (FEC) is presented, where the FEC scheme is based on Reed-Solomon (255,239) coding; the system performance is numerically evaluated and the dependence of FEC coding gain and transmission penalty on system parameters is investigated. The results show that FEC can greatly improve the system performance regardless of changes in system parameters.

We have reviewed some techniques which are used in supervision of EDFAs so far. Based on these techniques, we propose a new supervisory scheme and discuss some key problems in the scheme. This kind of SV system will be quite important for the all-optical networks in the future.

We have analyzed the relative intensity noise (RIN) of the pump lasers transferred to the signal in a Raman amplifier system using polarization-multiplexed pump lasers. The result shows that this RIN transfer could be reduced by 3 dB using the polarization-multiplexed pump lasers instead of single pump laser. In a Raman amplifier system, the pump modulation could be used for line monitoring. We have also analyzed the power penalty caused by the modulated pump laser in such a system.

All optical gain clamping technique of Raman fiber amplifier was experimentally investigated, and a better gain-clamping is achieved when clamping wavelength is in relatively low gain regime and multiwavelengths clamping technique is adopted. This dynamic gain control method is also suitable for the Raman amplifiers with multi-wavelength pump.

Raman amplification has emerged as one of the enabling technologies for high capacity and/or long-reach WDM networks. This paper will review the performance requirements of Raman pump modules in Raman amplified WDM optical networks. Pump laser parameters, such as relative intensity noise, linewidth, wavelength stabilization and multiplexing, degree of polarization, wavelength tunability, electrical power consumption, as well as their system implications will be discussed in details. Current progress on low noise, wavelength stabilized Raman pumps as well as high power tunable Raman pumps will be presented. Performance advantages and potential issues of Raman amplification in comparison to those of erbium doped fiber amplifiers will be summarized in terms of optical signal-tonoise ratio improvement, gain bandwidth, optical add-drop transients, network provisioning, and safety.

Raman amplification has been one of the enabling technologies employed to push the capacity x distance product in terrestrial DWDM systems. Due to its broad gain spectrum and the commercial availability of desired Raman pump wavelengths, more than 100-nm bandwidth is made possible covering the low loss window of silica fibers. In addition, distributed Raman amplification in transmission fibers and/or dispersion-compensating fibers (DCF) can significantly improve the system signal-to-noise ratio. In this talk, we will discuss the advantages and issues of Raman amplification, including pump-pump and pump-signal crosstalk, and their impact on the design of next generation fibers. Recent transmission experiments over Raman enabled fibers will be presented. Optimization of the system performance by allocating the Raman gains in co-, counter- and DCF-pumping will also be discussed.

In long haul wavelength division multiplexing (WDM) optical transmission systems, fiber Raman amplifier will become an important subsystem because of its unique advantages. This paper is concentrated on one kind of noise in fiber Raman amplifiers—amplified spontaneous emission (ASE) noise induced by parametric gain (PG). When a strong optical wave (pump) propagates along a fiber, it creates a spectral region around it where a small optical wave (noise) may experience gain. This phenomenon is known as parametric gain (PG). Although the noise generated by PG around signal wavelength is very small, Raman effects under high pump power can amplify it to a relatively high level. It interacts with the signal and affects the system performance. In this paper, PG induced ASE noise in the optical fiber Raman amplifier is analyzed and the related expressions are present, including single pumping and WDM pumping. The ASE noise in a 5 pumps, 64 channels and gain-flattened fiber Raman amplifier systems is also calculated.

Interleaver is one of the key components in the fiber optic communication systems. It provides a unique function of separating and combining two sets of WDM signals with very wide or very narrow channel spacing. This paper reviews the recent progress and future prospects of the interleaver technology, including birefringent interleaver, Gires-Tournois- Interferometer and all-fiber based interleaver technologies.

As a kind of optical comb filter, double-ring resonator formed by a 3 x 3 fiber directional coupler can be used in optical communication system. By changing the parameters such as the coupling angle of the directional coupler, physical lengths ofthe double rings, loss, and gain factors. It behaves like a device with multi-function involving multichannel band-pass and band-stop filter, laser resonator or amplifier, optical all-pass filter and so on. In this paper, characteristics of double ring resonator formed by an equal triangle 3 x 3 fiber directional coupler is numerically analyzed in detail, for the double-ring resonator with equal lengths, a minor peak appears near every main resonant peak, they change in reverse dynamically with the increase of the gain coefficient for fixed loss. For unequal lengths case, active cavity brings about clear Vemier effect, and free spectral range (FSR) is increased. At last, applications of this component in the construction of flattop interleaver and dispersion compensator in FDM system are discussed.

Three kinds of add-drop multiplexers (OADMs) based on fiber gratings and circulators are proposed and demonstrated. There are significant homodyne crosstalk reduction of about 24.4dB, 39.3dB and 23.6dB, 24.6dB on the dropped and added channels, respectively, for the best proposed lI-type structure and III-type structure as compared with the conventional I-type structure.

In the AWG device, a total internal reflection(TIR) mirror is fabricated at the middle ofeach arrayed waveguide based on silicon-on-insulator material. We simulate the field distribution of the output lightwave before and after a few TIR mirrors are damaged due to fabrication process and find that there is almost no any change of the output lightwave and the insertion loss ofthe AWG device, which indicates that fabrication ofthis type ofAWG device with TIR mirrors has high tolerance. Simulation results indicate that usage of tapered waveguides can reduce the insertion loss but affect the spectral response of the device. We design a rectangular groove in the middle of each output waveguide near the junction, of the output waveguide and the slab waveguide. Simulation results indicate that this type of structure can broaden the passband and improve the spectral response, while increase the insertion loss of the device a little in the same time.

We experimentally demonstrate a demultiplexing scheme for the frequency-interleaved dense wavelength division multiplexing (DWDM) radio-on-fiber system using an arrayed-waveguide grating (AWG). We first show the principle of the frequency-interleaved DWDM radio-on-fiber and possible configurations of multiplexing and demultiplexing (DEMUX) schemes using an AWG. In the experiment, 25-GHz separated 2-channel optical double sideband signals modulated by a 60-GHz millimeter-wave carrying a 156 Mbps data are optically multiplexed by frequency interleaving. The power penalty after DEMUX, which was due to interchannel interference, was less than 0.5 dB. We also made a transmission experiment in 25-km standard single mode fiber (SMF). No noticeable power penalty in the received data due to transmission was observed. This is because carrier and only one of the sidebands are detected in the proposed DEMUX scheme.

This paper describes the design method and performance of the optimized 10 Gbps uncooled DFB LD module. We discuss the optimal EOTM (electrical, optical, thermal, mechanical) design from the viewpoint of high speed OE(Opto Electronic) conversion between the electrical input and optical output. Total performance of the module is verified by a 10Gbps transmission experiment and a 10-Gbps fiber optic link over 20km of conventional SMF(single mode fiber) is successfully achieved. The module output power is +2dBm and ER(Extinction Ratio) is over 7dB at 10Gbps and 75°C.

Widely tunable laser has been recognized as one of the key enabling technologies for more efficient and intelligent optical networks. We present recent advanced development of a monolithic full band tunable laser device based on sampled grating distributed Bragg reflector (SG-DBR) technology at Agility Communications. We will discuss key performance parameters, as well as long term reliability of a widely tunable laser with high power (> 20mW), wide tuning range (the whole C-band, or L-band), monolithically integrated electro-absorption (EA) modulator and semiconductor optical amplifier (SOA). Optical network applications will also be discussed.

An actively harmonic mode locked fiber ring laser was setup, stability under the nearby of harmonic modulation frequency was analyzed. A new cavity length feedback system was implemented to realize the long-term stability. A chaos phenomena in the actively harmonic mode locked fiber ring laser was observed.

We demonstrate a novel approach for generating a narrow-linewidth and highly side-mode suppressed Erbium-doped fiber laser (EDFL) by combining it with an intracavity feedback-injected 1.55?m Fabry-Perot laser diode (FPLD) via optical couplers. To help selecting the strongest longitudinal mode from the gain-spectrum of FPLD for lasing in the EDF ring cavity, the intracavity FPLD is operated at just below threshold condition associated with an intracavity feedback optical power ratio of <1%. The lasing mode and center wavelength of the proposed EDFL source are decided by both the cross-correlated gain profile of EDFA and FPLD, however, the effect of FPLD injection modes is found to be more pronounced. The optimized lasing linewidth (system limitation) and side-mode suppression ratio of 0.01 nm and > 49 dB are obtained. The worst linewidth at 3-dB and 10-dB decay are observed to maintain at about 0.016 nm and 0.05 nm, respectively. Linear wavelength tuning of up to 6 nm (from 1551.5 nm to 1557.7 nm) by adjusting temperature of FPLD from 10 °C to 50 °C at just below threshold is reported. The wavelength-tuning slope is about 0.155 nm/°C under temperature accuracy of 0.1 °C .

The development of optoelectronic components for gigabit Ethernet communications is converging towards access networks where the cost of device makes a significant impact on the market acceptance. Device fabrication and packaging cost have to be brought down with novel assembly and packaging methods. Singapore has established a reputation in semiconductor device development and fabrication with excellent process and packaging facilities. Institute of Microelectronics (IME) was founded in 1991 to add value to the Singapore electronics industry. IME is involved in the development of active and passive photonics components using Silicon and polymer materials. We present a brief report on the development activities taking place in the field of optical component packaging at IME in recent years. We present a review of our competence and some of the optical device packaging activities that are being undertaken.

In this work, we review several different types of optical signal processors utilizing liquid-crystal technologies. These optical components include optical switches, voltage-controlled variable attenuators and optical harmonic equalizers for WDM (wavelength-division-multiplexing) applications.

A novel method based on combination of the bending and thermal characteristics of a new long-period fibre grating written on a photosensitive fibre by high-frequency CO2 laser pulses is proposed for the gain flattening of EDFAs. The gain profile was flattened to within ±0.7dB over a 32nm bandwidth.

An L-band Erbium-doped fibre amplifier consisting of a length of dual-core fibre is reported. It has a flat gain of 12.1dB from 1565nm to 1615nm with a maximum gain variation of 0.6dB, and noise figure less than 3.9dB. We have a good corroboration between the theoretical simulations and the experimental results to show that the coupling coefficient between two cores can be increased with over the wavelength range 1550 nm to 1620 nm by adjusting the core-to-core spacing and core parameters to compensate for the gain decay due to erbium alone greater than the wavelength 1550 nm.

Dynamic EDFA controlled by pump compensation has been successfully developed with transient suppressions within 1dB, and average within 0.3-0.45dB in various kinds of channel loading configurations, gains, and input powers for metro networking applications.

A novel tunable gain equalizer based on the LPFG induced by high-frequency CO2 laser pulses is demonstrated, for the first time, to our knowledge. The resonant wavelength and the resonant peak amplitude of the novel LPFG have a good linearity with temperature and transverse load, respectively. Therefore, the wavelength and the peak amplitude of the gain equalizer can be tuned by the means of adjusting temperature and transverse load independently, showing that this tunable equalizer is a practical approach to realize dynamic gain equalization in EDFA systems.

Due to stimulated Raman scattering, shorter wavelength channels will pump longer wavelength channels, and so transfer their energy to longer wavelength channels. A tilt occurs when a flat DWDM signals travel through transmission fiber. We characterized the Raman tilt for 100 km SSMF and LEAF fiber at composite input powers ranging from 13 dBm to 22 dBm with a 40-channel laser source with 100 GHz spacing. Raman tilt with different fiber length at different composite launch power for SSMF is fully studied. In order to compensate the positive Raman gain tilt generated in the transmission fiber, a negative tilt is required. A tilt control function is available in some EDFAs. The mechanism of this tilt control is to use a tunable attenuator to change the internal loss of the amplifier. Unfortunately, the negatively tilted gain spectrum achieved by this mechanism is not a straight line. From the simulation result, the tilted gain curve actually can be regarded as two straight lines with a junction at about 1538 nm in good approximation. By combining the Raman output spectrum with the EDFA output spectrum, tilt is eliminated, but a non-flat spectral shape is generated with a dip at 1538 nm. This combined spectral shape agrees quite well with the measured spectral shape in actual system. This study reveals imperfection with this tilt compensation mechanism. A new tilt compensation solution is proposed and tested. Test results shown significant flatness improvement.

In this paper we will present ultrafast all-optical clock recovery up to 100 Gb/s using two-section gain-coupled DFB lasers. Both the coherent and incoherent schemes will be described. The two-section DFB laser can be used in combination with semiconductor optical amplifier so that clock recovery for both the RZ and NRZ formats can be achieved. For the RZ format, the sensitivity of clock recovery is 50 ?W, the jitter is less than 1 ps at 40 Gb/s. We use parametric amplification for all-optical regeneration. Extinction ratio improvement over 11 dB has been achieved.

The performance improvement of 40Gbit /s soliton transmission system employing nonlinear optical loop mirror (NOLM) as all-optical regenerative repeater is investigated. The switching characteristic of NOLM is optimized using our design strategies. By inserting the optical repeater where the eye-penalty of the transmitted signal becomes 1dB, we achieve the improvement of 0.3dB eye-penalty resulting in further 480km transmission distance.

A passively Mode locked soliton fiber ring laser was successfully demonstrated, and transform limited soliton pulse with FWHM width of 158 femto second was achieved.

We demonstrate experimentally a novel form of double-pulse solitary waves in a passively mode-locked fiber ring laser. The new solitons are characterized as that they consist oftwo single solitons with fixed, discrete pulse separations. Through studying the interaction between bound-soliton pairs observed in a passively mode4ocked fiber laser, we firstly revealed that the bound soliton pair behaves as a unit, and has exactly the same interaction properties as those of the single pulse soliton in the laser. Numerical simulations confirm the existence of the new form of solitons in the laser.

The effect of time-delay-ripples on 40Gbit/s filter cascade system and multispan transmission system is investigated for the first time. It is shown that ripple amplitude basically determine EOP in the practical transmission system.

We fabricated a low-power CMOS post amplifier for asynchronous transfer mode passive optical network system. In this amplifier the multistage feedforward type was employed for high-speed response. The circuit consists of adaptive threshold control with wide input range and limiting amplifier. This amplifier was fabricated in a standard 3.3V 0.35?m CMOS technology and showed a low power consumption of 80mW at the supply voltage of 3.3V. The limited output voltage swing was 200mVP-P .

Facing the new demands of the optical fiber communications market, almost all the performance and reliability of optical network system are dependent on the qualification of the fiber optics components. So, how to comply with the system requirements, the Telcordia / Bellcore reliability and high-power testing has become the key issue for the fiber optics components manufacturers. The qualification of Telcordia / Bellcore reliability or high-power testing is a crucial issue for the manufacturers. It is relating to who is the outstanding one in the intense competition market. These testing also need maintenances and optimizations. Now, work on the reliability and high-power testing have become the new demands in the market. The way is needed to get the ‘Triple-Win’ goal expected by the component-makers, the reliability-testers and the system-users. To those who are meeting practical problems for the testing, there are following seven topics that deal with how to shoot the common mistakes to perform qualify reliability and high-power testing: • Qualification maintenance requirements for the reliability testing • Lots control for preparing the reliability testing • Sampling select per the reliability testing • Interim measurements during the reliability testing • Basic referencing factors relating to the high-power testing • Necessity of re-qualification testing for the changing of producing • Understanding the similarity for product family by the definitions

Device performance of Ti:LiNbO3 10Gbps intensity modulator and 5Gbps phase modulator has been measured over various temperature in terms of insertion loss and bias drift voltage. Insertion loss and DC bias drift changed by 0.4dB and 0.45V, respectively, whether RF&DC port is open or short.

The critical aspects of signal sampling in the context of optical measurements are reviewed. A parallel is drawn between optical- and time-signal sampling, which shed light on the impact and interplay between key parameters. All key parameters have their mirror image in both pictures, time ? frequency or optical-frequency ? optical-delay. The equivalence is made specific, between optical sampling-step, resolution bandwidth, interferograms, photodetector bandwidth or laser lineshape on one hand, and sampling rate, low-pass filter, signal spectrum and signal bandwidth on the other hand, along with the conditions they should fulfill, in both pictures, given an actual setup.

In this paper, a novel micro pseudo-spherical lens is proposed to resolve the intrinsic high insertion loss problem associated with the micro spherical lens, and yet retain its other good qualities. The design parameters for this micro pseudo-spherical lens are optimized using Gaussian beam matrix propagation method. A geometric ray-tracing program is also used to calculate the coupling efficiency for collimator pairs. The simulation results for all four different types of micro lenses (pseudo-spherical, spherical, aspheric and GRIN lenses) are presented and compared to experimental results. It is shown that the novel micro pseudo-spherical lens presents significant performance improvement over the micro spherical lens, and offer similar performance to GRIN and micro aspheric lenses.

For 10 Gb/s and 40 Gb/s NRZ signals the maximum reach of a WDM transmission system depends on the optimization of the inline dispersion scheme, which is different for both data rates for SPM limited transmission systems. Therefore, a mixture of both data rates in the same fiber may cause an impairment of the system performance. This can be reduced introducing additional pre-compensation. The value for the optimum precompensation varies with the inline dispersion scheme and the launch power. A design rule to determine the optimum pre-compensation was derived empirically for SSMF and NZDSF. For fixed inline dispersion compensation on SSMF, the maximum reduction of the system reach without pre-compensation is found to be up to 6 dB, compared to optimized inline dispersion compensation schemes. By applying optimal pre-compensation this impairment can be reduced to less than 1 dB. For NZDSF with pre-compensation the system performance can also be improved significantly. Thus WDM systems using different line rates at the individual wavelengths are possible with only a small penalty.

We report on the 320-Gb/s (40 Gb/s x 8 channels) WDM transmission experiment using short-period dispersionmanaged fiber. After 320-km transmission, the average Q-factor was measured to be better than 20.4 dB. In addition, we compared the performances of various types of fibers in 40-Gb/s based systems.

When closely temporally separated two optical pulses of identical wavelength are transmitted in optical fiber, the pulse separation could be changed due to nonlinear effect. We evaluate the pulse-to-pulse interaction in dense dispersion management by simulation and experiment. First, we evaluate the influence of third order dispersion on the pulse-topulse interaction by numerical simulation of transmission. The interaction can be effectively reduced when dispersion shifted fiber for dispersion slope tuning is placed at input of EDFA repeaters. Then we performed a recirculating loop transmission experiment of 3.5-ps optical RZ pulses with 12.5-ps initial separation in dispersion flattened dispersionmanaged fiber. The interaction can be greatly reduced by employing dense dispersion managed fiber, which is favorable for ultra-high speed OTDM transmission systems.

The chromatic dispersion of optical fibers degrades the transmission characteristics of optical communication systems. In high-speed WDM systems both dispersion compensation and higher order dispersion (dispersion slope) compensation are necessary. In this paper, a new dispersion slope compensator based on tapered fiber gratings is proposed. The dispersion slope compensator consists of two linearly tapered fibers with uniform Bragg gratings.

The evolving technology of high-order-mode dispersion management is discussed. Selected high-order modes of specially designed fibers can have high dispersion ofa few hundreds psec/nm*km, controlled dispersion-slope in the range of 1.5 to 10 psec/nm2*km, and large effective area of6O to 8Ojim2. Thus, the technology enables the construction of dispersion management devices with low-loss, accurate broadband dispersion compensation for a variety of transmission fibers, and low sensitivity to non-linear effects. Compared to standard fundamental-mode-based devices, the high-order-mode-based devices may have as much as 6 dB advantage in insertion loss and 6 to 13 dB advantage in power tolerance to non-linear effects. Specific high power tests show a 13 dB higher threshold to stimulated Brillouin scattering and 6 to 10 dB advantage in sensitivity to self-phase modulation. Line amplifiers designed and built around these high-order-mode devices may have an advantage of I dB or more in Noise Figure, thereby enabling significantly longer reach ofoptical links.

As bit-rate in optical fiber communications system increases to 4OGbit/s or beyond in single channel, polarization mode dispersion (PMD), especially higher-order PMD, has become one of major obstacles in high-speed and long-distance transmission systems. Therefore, it is obvious that there are two ways to think of higher-order PMD, one is to investigate that how the higher-orderPMD affect system performance and its critical limitation value; the other is to find effective methods to mitigate or compensate higher-order PMD to improve system performance. Therefore, in this paper, firstly a detailed discussion will focus on the system impairment due to higher-order PMD, secondly, a improved first-order PMD compensation method is proposed to compensate part of higher-order PMD and checked with numerical simulation.

Based on the modeling of considering polarization mode dispersion (PMD) and polarization-dependent loss (PDL) simultaneously, the impact ofPDL on high-speed PMD compensation systems is discussed in this paper. The simulation result shows that for the 4OGbit/s NRZ systems being PMD value 8ps, if the PDL value of single device increases to 0.7dB, the infkience of PDL on the systems performance is obvious. Furthermore, if it reaches to 1.5dB, the outage probability of eye-opening penalty of PMD compensation system will increase from 10-6 (in case no PDL effect) to 10-4.Therefore, it is necessary to considering the impact of PDL on systems performance in the PMD compensation systems design ifthe PDL value ofdevice reaches to 0.7dB or beyond

Based on the segmented bi-refringent fiber model, the penalties due to PMD, GVD and nonlinearity effects in 4x400bit/s WDM systems is calculated by solving the coupled nonlinear Schrodinger equation in bi-refringent fiber with split-step Fourier method for different PMD values in this paper. The simulation results show that in a WDM system the outage probability for one channel is not much larger than that in other channels. For PMD compensation in WDM system it runs a dangerous risk to only compensate the worst channel at a time, and it might be necessary to compensate all channels simultaneously.

This paper presents a method of splitting signals from fast and slow axes of polarization maintaining fibers by using a polarization splitter. The new scheme includes a turning connector and a polarization splitter. The polarization splitter will split signals from a fiber and give two separated outputs. It gives some theoretical calculations and the relations between the turning angle and output power from the splitter. It also analyzes the relationship of electrical power spectrum versus differential group delay and relative angles between the axis of a fiber and direction of a polarization splitter. Preliminary experiments are also given in this paper. The scheme has a great potential of becoming an effective method of solving polarization mode dispersion (PMD) problem.

We report on the effect of the polarization-dependent loss (PDL) on the polarization-shift-keying (PolSK) transmission system. In a PolSK system, the PDL of the fiber-optic link could generate an inband crosstalk and degrade the signal’s extinction ratio. The result shows that a 3-dB PDL could result in the 1-dB power penalty. However, unlike the effect of PMD, it is not dependent on the transmission speed.

We present the methodology and the results from field measurements along installed fiber links on the localization and classification of fiber sections with respect to their polarization-mode dispersion (PMD) characteristics using a polarization sensitive OTDR (P-OTDR). The technique relies on the statistical analysis of the local degree of polarization (DOP) obtained from the backscattered signal. It allows classifying fiber sections as low-, medium- and high-PMD contributors.

An optical liquid drop sensor(OLDS) which can be used to distinguish different liquids is presented in this paper. The main block ofthis system are a light path and a drop-forming head sensor. We can gain three kinds ofiniormation about the liquid to be measured with this OLDS system. That is optical signal representing light intensity information, electrical signal revealing volume information ofthe liquid drop and spectral signal displaying ingredient information of the liquid individually. Colligate this three kinds of information and after data-processing the fingerprint of the measured liquid can be derived. Due to different characteristics of various liquids, the fingerprint of one liquid differs from that of others. This difference has been proved through experiment. Thus the OLDS is valuable for the discrimination of different liquids and the measurement ofthe liquid's parameters.

The effect of diametric load on the wavelength of the reflected light from a fiber Bragg grating (FBG) as a transverse load sensor is investigated experimentally. The diametric load will cause the stress-induced birefringence in optical fiber. This birefringence will separate the reflected resonance peak of the fiber Bragg grating. The spacing of the two separated reflected peaks indicates the diametric stress, which is proportional to the transverse load on the fiber. In this paper, we present some experimental results of transverse load sensing using a fiber Bragg grating. An erbium doped fiber (EDF) amplified spontaneous emission (ASE) source is introduced as a broadband light source, and the reflected light from the grating is directly into an optical spectrum analyzer (OSA) to observe the reflected peaks of the Bragg grating. In our experiment, the transverse load is from 0 to about 2.3N/mm. The resonance peak separation is very distinct when the load is higher than 0.8N/mm. We also expand the sensor into multi-point distribution sensing. Three FBG’s with different resonance wavelength are placed in the sensing circuit. This experiment system can be applied in transverse load sensing, especially in smart structure or precision structure in which the size of the sensor is limited strictly.

Elongation characteristics of an optical fiber loop reflectors are investigated theoretically and experimentally considering fiber birefringence of the loop. Due to the birefringence formed from elliptical deformation and/or unsymmetrical stress (for example, bending) and/or twisting, the reflectivity and elongation characteristics would have some changes. During elongating process of drawing a fused taped coupler, the transmitted power of a fiber loop reflector showed some interesting characteristics as we were twisting the fiber. Even we did not twist the fiber, due to the elliptical deformation and unsymmetrical stress, the envelope of the elongation curve displayed some unsymmetrical modulation characteristics. These characteristics were more evident when the length of the fiber was longer. Using coupled-mode theory, all these experimental results were well explained.

The optical signal along the light-path in the optical network will traverse a cascade of optical filters. The equivalent pass-band ofthe cascaded filters decreases in comparison with a single filter, and it is required that the light sources should be much more stable. In this paper, we investigated the cascaded characteristics of FILM filters that are widely used in WDM systems due to their flat pass-band and low insertion loss. Both the influence of amplitude response and the influence of phase response are considered together. Firstly, we give two equations of the pass-band characters and the phase response of FILM filters, and then calculate the variation of the cascaded FILM filters' bandwidth versus the number of the cascaded filters. We evaluated the influence of the filters' phase nonlinearity in more detail. The simulation results imply that the frequency range ofthe cascaded FILM filters under the condition of zero power penalty decreases, and the power penalty curve caused by phase nonlinearity fluctuates more drastically than that without the consideration ofthe phase response.

We have studied theoretically the formation of short pulses inside an actively mode-locked ring laser by studying the effect of pump power, amplification parameter and modulator characteristics. Moreover, multi-soliton formation has been examined.

A new scheme is put forward for generation of the compressed pulse pair using fiber grating. This new compressor bases on the soliton-effect and makes use of high-order soliton supported by the fiber grating. In the anomalous GVD regime outside the photonic band gap ( ? < ? B), the third-order soliton may be formed when pulse pass through the fiber grating. As we know, the third-order soliton will split into two distinct pulses at z0/2(zois the soliton period), but the two pulses recover the original shape at the end of the soliton period. If we break the balance between the GVD and the nonlinearity at z0/2, the split two pulses will be self-existent permanently but not recover. We present the basic idea and introduce this new kind of compressor in this paper. Finally, we have proved this method by numerical simulation with SSF (Slip-Step Fourier) method.

On the basis of the arrayed waveguide grating (AWG) and the optical fiber grating, a novel kind of dispersion compensation used for fiber communication is presented in this paper. First, a dispersion equation of a uniform fiber grating in the vicinity of its reflection band gap is deduced using the mode-coupled theory. Then, a dispersion compensation of a wavelength division multiplexing (WDM) system is analyzed using the AWG and the uniform fiber grating. Finally, the results of dispersion compensation for an 8-channel WDM system are calculated and discussed. It results in that this method can compensate the dispersion of every channel of the WDM system respectively, and make the compensation efficiency maximum theoretically.

In future high bit-rate DWDM systems, the channel separations will be narrower and the bit-rate of single channel will be higher. The dispersion in transmission of fiber Bragg grating is a factor that cannot be neglected. We emphasize the effects of the 3rd order dispersion because the 3rd order dispersion is more detrimental for short pulses than the 2" order dispersion. Furthermore, the 3rd order dispersion is always positive whether the wavelength of the adjacent channel is longer than Bragg wavelength or not. So the accumulated effects are more notable than the 2nd order dispersion, which can be positive or negative. We calculated the effects of the dispersion of FBG on pulses with ips, 2ps and 5ps pulse widths under channel separations of 5O GHz and 100 GHz. It is shown the 3rd order dispersion will be more detrimental in high bit-rate DWDM systems. This paper will give some guide to FBG design for ADM application. The method we use is split-step Fourier method.

In this paper, we present a novel method for the fabrication of an EDFA gain flattening filter(GFF) with two identical chirped fiber gratings formed in series after a double exposure. This method shows a significant improvement in the flattening level and error function in GFF, compared with the conventional method for forming the OFF by using single chirped fiber grating for equalizing the gain spectrum.

In this letter, the gain characteristics of the erbium-doped fiber ring amplifier at below laser threshold using a I :99 coupler, which feed back 1% ofthe output power into the ring cavity, has been studied. In the same condition, the gain of the erbium-doped fiber line amplifier has been measured, which is lower 3.2dB than that of the erbium-doped fiber ring amplifier. In the region of 1520-1570nm, the gain ofthe erbium-doped fiber ring amplifier is bigger 1.5-2.5dB than that oferbium-doped fiber line amplifier.

A group of equations deduced from thulium-doped fiber amplifier (TDFA) rate equations and light propagation equations, introducing the overlapping factors between the (pump and signal) light intensities and the thulium dopant distributions inside the fiber core, is presented in this paper. Using the group of equations, the small signal gain and optimum fiber length and threshold pump power characteristics ofthe amplifier are simulated.

In this paper, a novel design method used for the birefringent interleaver is presented, which is based on Harris systematic design theory' with some improvements in respects of the approximation of the number of crystal used and the dispersion compensation of crystal materials. In our method, as optimal designing, the required frequency response ofinterleaver as the target function is specified in first step ofcalculation, and some basic design parameters such as the number of crystals and rotation angles of each element will be automatically obtained. This design method is flexible for different crystal materials and is extensible for the consideration ofthe manufacture tolerances and the temperature compensation ofcrystal. As a design example, simulation results are presented and relative analyses are taken.

Optimal dispersion compensation ratio (DCR) shall be carefully chosen for different dispersion-managed systems. We simulated the dispersion-managed systems of 40 Gb/s bit rate and present the optimal values for different signal powers and transmission lengths.

The development of Optoelectronic components for communications is converging towards access networks where device cost makes a significant impact on the market acceptance. Thus, the device design engineer needs to input assembly, fabrication and process constraints into the design at an early stage. The present study is part of a Project on Packaging of Optical Components that IME, Singapore has initiated as part of an ongoing Electronics Packaging Research Consortium with industry partnership. In the present study, the coupling of optical radiation from a laser diode to optical fiber is simulated for a fiber optic transmitter component development project. Different optical configurations based on direct coupling, spherical ball lenses, integral lensed fibers and thermally expanded fibers are created within the commercially available transmitter package space. The effect of optical element variables on the placement tolerance is analyzed and will be reported. The effect of alignment tolerances on the optical coupling is analyzed. Simulation results are presented recommending realizable alignment and placement tolerances to develop a low cost short range link distance transmitter.

We numerically analyze and compare effect of polarization mode dispersion on an optimal 40Gb/s RZ bi-directionally pumped all-Raman and EDFA transmission system. Although bit-error-ratio distribution is similar for both systems, Raman amplified transmission system shows a higher tolerance to polarization mode dispersion due to a relatively lower bit-errorratio base induced only by nonlinear effects.

In this paper, gain and noise characteristics of gain-clamped Fabry-Perot lasing erbium-doped fiber amplifier (EDFA) is investigated in detail when reflective wavelength and reflectivity of the fiber Bragg gratings (FBG) are selected different values. The relationships between gain and noise performance of the gain-clamped EDFA and the selection ofFBG parameters such as reflectivity and reflective wavelength are investigated and analyzed. It is found via numerical simulation that the selection of FBG parameters has important influence on the performance of the EDFA. The results obtained in this paper can provide significant datum and reference for optimizing and improving the performance of all-optical gain-clamped EDFA.

Using the feature that the gain peak wavelength only depends on the pump wavelength, the multi-pump sources and their gain peak wavelengths can be considered as the input and output variables of the single layer feed-forward neural networks, and the optimal design for multi-pumped Raman Fiber Amplifier can be converted to a optimal problem of the single layer feed-forward neural networks. With the optimal algorithm the Raman gain spectra of multi-pumped FRAs were investigated. The results show that the optimal configuring algorithm is simple and effective.

A bulk design of optical interleaver with 50 GHz free spectral range (FSR) is experimentally demonstrated. The optical interleaver consists of three cascaded stages of modified Mach-Zehnder interferometers (MZIs), which are fabricated with beam splitters, end-mirror and delay-blocks. Beam splitters are used to split a propagating beam into two beams or coherently recombine two split beams, and delay-blocks located in one arm of each MZI are used to control the phase difference. The interleaver processes square-like spectral response with nearly zero ripple, wide flat-top, and rather low crosstalk, sharp steepness, and ease in fabrication as well. The bandwidth of passband at -0.5 dB and stopband at -15 dB are over 30 GHz and about 25 GHz, respectively. The insertion loss is smaller than 1.5 dB and the polarization dependent loss (PDL) is about 0.15 dB. The crosstalk is lower than -25 dB. The return loss is larger than 45 dB.

A cascaded M-Z fiber interferometer has been analyzed,which is made of several 3dB couplers in series. Theoretical analysis and numerical simulation indicate that a good comb-like curve will be obtained when the length difference ? L of interference arms is given a certain value in order. Side-mode suppression ratio (SSR) will increase as adding of the number of interference arms.

We demonstrate a high-gain, low-noise in-line Er-doped optical fiber amplifier (EDFA) using a novel long period fibre grating written by high-frequency CO2 laser pulses. The optimization of such an EDFA has been investigated, for the first time, to our knowledge.

A formula used to calculate the power penalty caused by optical nodes such as OADM/OXC nodes in WDM all optical networks is deduced on the base of interference between electric fields by taking the phase character of node transmission into account. To prove its correctness, we use the formula to figure out the power penalties in the case where the central frequencies of multiplexer/demultiplexer (MUX/DEMUX) stand away from the ITU-T grid, and compare them with experimental results. Both calculating and experimental results show that when central frequency of MUX/DEMUX changes from —30GHz to 40 GHz with respect to ITU-T grid, the corresponding power penalty changes up to about 6.5dB, and the maximum difference between them is 0.5dB.

Absence of relaxation oscillation in SOA fiber ring lasers is analyzed in this paper, which is always presented in actively mode- locked erbium-doped fiber (EDF) ring lasers. The super-mode competition can be effectively suppressed in SOA ring lasers due to the gain saturation of SOA. The experimental demonstrations were confirmed the above issues and good stability in a SOA mode-locked fiber ring laser without any feedback circuits was shown.

The system approach considers a priori the interdependence of all technologies integrated into a MEMS product. Parallel development of these technologies and enabling wafer-level batch manufacturing through process integration is key to successful product development and meaningful cost reduction.

In this paper, a new kind ofhigh precision mechanical variable optical attenuator (VOA) based on optical blocking slice is designed, fabricated and tested. The attenuation range is more than 40dB. The measured insertion loss and return loss are less than 0.5 dB and larger than 65dB respectively. The best resolution is 0.001dB. The attenuator has high attenuating precision, low insertion loss, high return loss, polarization independent, wide temperature range, wide spectrum scope, wide attenuation range, low cost and so on.

Wavelength switching is demonstrated in an actively mode-locked fiber laser by using an F-P laser diode as a modulator and a HiBi fiber loop mirror as a periodic multichannel wavelength filter. By controlling the temperature of the laser diode, wavelength switching in a range ofabout 10nm was achieved.

A novel Interleaver using a modified Michelson interferometer in which its reflecting mirrors are replaced by three-mirror Fabry-Perot interferometer and Gires-Tournois resonator respectively is presented. On basis of the theory of Michelson interference, a designing principle of the interleaver, which has square-like frequency response with zero ripple, wide flat-top, unity contrast, simple structure, and stable performance, is given. The device which possesses channel spacing of 50GHz, zero ripple, 0.5dB bandwidth more than O.27nm, and close channel isolation more than 73dB has been designed.

DWDM thin film filter phase properties including phase shift and group delay are analyzed in this paper. According to the dispersion control requirement, a method of group delay ripple compensation on transmission is forwarded. A high reflective film system is overlapped before the transmission light from the filter; the total group delay ripple is decrease with the counteraction ofhigh reflective films on contrast to the DWDM filter.

Based on the theory of polarized light interference and multibeam interference, the transmission function of BGT Interleaver is provided in this paper. By theoretical analysis of the transmission function, it is found that higher reflectivity of G-T cavity means broader interleaver pass and isolation bandwidth and smaller contrast. Otherwise, the narrower the pass and isolation bandwidth and the largger contrast. The related experiments are designed and demonstrated to inspect BGT interleaver characteristics, from which restriction relationship is reached between bandwidth and contrast of BGT interleaver. The higher reflectivity of G-T cavity coresponses to better bandwidth and worse contrast. BGT Interleaver can be designed according to above.

The repetition rate multiplication based on the temporal fractional Talbot effect is one of the solutions to realize the high-repetition frequency optical pulse. In this paper, varying even-multiplications due to the theoretical different group velocity dispersions (GYDs) of the linearly chirped fiber Bragg grating (LCFBG) are calculated by another way. The characteristics of repetition-rate-multiplied pulse tram caused by different values of deviation of GVD are discussed. And the range within which the deviation of GVD must be limited during the fabricating of LCFBG for the case of M=4 is 5% when the FWHM of the original pulse train is 12 PS. The cascaded multiplication of repetition-rate using the same or different GVD LCFBGs are also theoretically explained.

These chirped fiber Bragg grating is an attractive and practical element in particular context of dense wavelength-division multiplexing. We introduce a structure to realize working mode of the optical pulse's high power amplification. Also we analyze the variation of pulse signal after broaden or compression together with the variation in frequency domain and the peak intensity of pulse signal amplified by EDFA and subsequently compressed, in which these influence caused by the discrepancy of two optical paths and two dispersion parameters are considered.

This paper proposes a novel polymer optical fiber coupler, which use thin optical fiber bundle to split and combine the light between thick POFs. The advantages of the coupler over mixing rod coupler include reduced coupling loss, elimination of fiber bending loss, and flexible in design of splitting ration and number of ports. The construction of the coupler is described in detail. The design ofthe thin optical fiber bundle and the relationship between quantity of fibers in bundle and packing fraction are analyzed. 31-fibers bundle is found to be appropriate for constructing SBC. A directional coupler and a 4 X 4 star coupler are constructed as examples.

A novel double-grating design of widely tunable Ytterbium double-clad fiber (DCF) laser is demonstrated. By using an extermal cavity in double-grating configuration the laser wavelength is tunable over a range of 72 nm from 1037 to I I 09 nm. And a spectral line width of GHz over the whole tuning range is achieved. Measurements indicate the laser is linear polarized.

After taking into account the wavelength-depended reflectivity distribution profile of a fiber Bragg grating, the oscillation wavelength ?l of fiber Bragg grating semiconductor lasers has been investigated theoretically. The results show that the laser oscillation wavelength ?l is not fixed at the Bragg reflection wavelength ?B of fiber Bragg grating, and the offset between ?B and ?l depends on the reflectivity distribution profile of fiber Bragg grating and the gain profile of semiconductor gain medium.

An interleaving filter based on crystal waveplate (Yttrium Vanadate) is presented in the paper. The properties of its output spectra are analyzed through Fourier expansion. Numerical calculation shows that one stage of crystal waveplate only gives a cosine shape power transfer function, which is not able to meet the demand ofchannel passband in system. In order to get flattened transmission and rejection band, two stages of crystal waveplate of a special angle between optics axes is used. Length of the second stage waveplate is as two times as the first one's, and it brings a thrice harmonious in power transfer function which can improve the spectra to get a box-like frequency response. By use of Fourier expansion the performance of the output spectra is optimized to meet the demand of DWDM systems. Experimental results are in agreement with the theoretical analysis well.

We investigated the window properties of SOALOM (Semiconductor optical amplifier in a loop mirror) used as a 3R regenerator using simulation method. A multi-section model of SOA (semiconductor optical amplifier) was used in our simulation. The switching pulse was assumed to be a noisy Gaussian pulse with finite ER (Extinction Ratio). The influence of some system parameters such as the pulse width and peak power of the switching pulse, the applied current to SOA, and SOA displacement in the loop on the window properties was discussed.

A procedure for designing a polymeric AWG is presented by considering the main required specifications. An optimal design is obtained for a 8 x 8 polymeric AWG operating around 1.55 µm witha channel spacing ofO.8nm. The design is verified by a simulation with a beam propagation method. For the designed AWG, the crosstalk (for adjacent channels) is well below —40dB and the insertion loss is about 1.6dB and the loss non-uniformity is less than 1dB. Effects of some fabrication errors on the performances ofthe designed AWG are discussed.

The repetition-rate multiplication of a pulse train outside the laser cavity by using all-pass filters is discussed. Schemes based on the phase responses of optical all-pass filters for multiplying the repetition-rate of an optical pulse train are presented and the design of the required all-pass filters is also presented. These methods, in principle, can produce a uniform pulse train with a multiplied pulse repetition-rate without energy loss. The error analyses of these methods are discussed and the error ranges ofthe parameters are given.

The general equations of transmission spectrum, reflection spectrum, and phase of a multi-minor F-P interferometer are derived. In view of the theory of Fabry-Perot interferometer tuning phase, a one-cavity F-P dispersion compensator is designed, and chromatic dispersion property of it is analyzed. Furthermore, the best designing parameters are obtained, and an improving scheme, which is practical and effective, is put forward.

A new optimization method for the design of integrated band-pass traveling-wave modulator based on aperiodic domain-inverted grating is demonstrated. The frequency response of 10GHz and 20GHz (0.5dB) centered at 200GHz are presented. The physical mechanism for this kind of modulator is ascribed to the constructive interference effect of the overall domains. The sequences and the length of the domains are optimized to realize the pre-designed wide, flat frequency response over the required frequency range by use of the simulated annealing (SA) method.

A simple and low-cost solution for interleaver used in optical communication, based on multi-cavity Fabry-Perot optical thin film filters, is presented. The analyses on this kind optical interleaver reveal the influencing factors that would affect the thin film interleaver performance, including the cavity's spacer thickness, ambient temperature variation, dispersion for the refractive index of deposited material depending on wavelength and the numbers of cavity that is important for passband flatness and isolation. The interleaver achieved to 100GHz-spaced channels, at wavelength around C band, with adjacent channel's cross-stalk less than —25dB and the minimum insertion loss nearly 0.4dB or so.

We investigate theoretically the process of supercontinuum (SC) generation which results from four-wave mixing (FWM) taking into account the effects of self-phase modulation (5PM) and cross-phase modulation (XPM) in a dispersion-shift optical fiber. The optimum pump wavelength at the highestconversion efficiency of FWM is obtained under the condition of two pumps with a weak probe on a lossless dispersion-shifted fiber and three pumps with equal channel spacing on a fiber including fiber loss. We also investigate the effects of 5PM and XPM by the spectra of co-propagating pump pulses with different pump power and fiber length. Finally, we propose a way to obtain the optimal SC resulting in a highest efficiency of FWM with small effects of 5PM and XPM.