Optical packet switching is commonly considered as a possible technology for future telecommunication networks, due to its compatibility with bursty traffic, eg Internet protocol (IP), and efficient use of wavelength channels. Current transport networks are voice-optimised and connection oriented, however the amount of data traffic is rapidly increasing, resulting in a continuous increase of average traffic through major exchanges exceeding 30% per annum (in Europe). Thus optical packet switching is seen as a future technology that will support diverse traffic profiles and give more efficient bandwidth utilisation through its ability to provide multiplexing at the packet level. In recent years the significance of optical packet switching as an emerging technology has been identified and researched by a number of research groups. Earlier optical packet switching demonstrators presented switching of mainly ATM compatible synchronously transmitted packets at bit rates up to 2.5b/s with the optical header encoded either in series or in parallel to the payload using the sub-carrier modulation technique. More recent projects have demonstrated switching capabilities at 10Gb/s using more advanced approaches with special encoding schemes for header and header detection, together with sophisticated control mechanisms for contention resolution. The capability of switching optical packets at bit rates up to 160Gb/s has recently been demonstrated. This paper discusses the architectures currently proposed for high speed optical packet switching, including the key techniques of header processing and payload switching. The focus is on a high speed demonstrator [OPSnet] capable of operation at rates >100 Gb/s.

A kind of optical coarse packet switching based on optical label routing is intrduced. In the switching, the swapping granularity is coarse compared to common optical packet switching, and the header of optical packet is labeled by multi-wavelengths. Some realization techniques on core switching node and edge node with terminator are studied. A principle experiment system is established to perform the optical coarse packet transmitting, switching, and receiving, and verify the feasibility of such optical coarse packet switching.

An adaptive double-stage PMD compensator capable of compensating first- and second-order PMD simultaneously is experimentally researched by introducing genetic algorithm for the first time. In order to verify the effect of the double-stage PMD compensator, an experimental optical transmission system at a bit rate of 10 Gbit/s is setup. In experiments, degree of polarization (DOP) of received optical signal is adopted for PMD monitoring. Experimental results show that genetic algorithm is effective and powerful in first- and second-PMD compensation. The DOP of received optical signals after PMD compensation can reach a value larger than 0.97 within 180 milliseconds during which genetic algorithm runs 50 iterations. PMD measurement indicates that the first- and second-order PMD in transmission link can simultaneously reach the minima after PMD compensation at the operation wavelength.

In today's high-bit-rate WDM systems, it is essential to monitor the residual chromatic dispersion (CD) to ensure that it does not exceed the designed tolerance. Among the schemes for CD monitoring reported so far, inband subcarrier tone method is relatively simple and effective for CD monitoring. However, this technique may be influenced by both polarization mode dispersion (PMD) and the chirp fluctuation of the external modulator. In this paper, we investigate the effect of PMD and chirp on CD monitoring and show that the presence of PMD and chirp induces significant CD monitoring errors. To tackle this problem, we propose a CD monitoring technique to suppress the influence caused by PMD and chirp fluctuation. In the proposed CD monitoring scheme, two RF tones are added at the transmitter. The light is coupled into an apriori known dispersion offset and then split into two branches in the monitoring module. A fiber Bragg grating filter which can remove one sideband is inserted before the photodetector of one branch. The PMD effect is eliminated by optically sideband filtering and RF power ratio detection, and the monitoring error induced by the small chirp fluctuation can be suppressed using two RF tones and a dispersion offset. The operational principle is analyzed and the experimental investigation is presented. Experimental results show that this technique can accurately monitor the accumulated CD without being affected by the PMD and small chirp fluctuation.

We demonstrated an adaptive polarization mode dispersion (PMD) compensation in one channel of a 40 channel dense wavelength division multiplexing (DWDM) transmission system with 40Gbps carrier suppressed return to zero (CS-RZ) modulation format.

We experimentally studied the PMD mitigation schemes using two different optical 3R regenerators. One is based on synchronous modulation and the other using an optical decision gate based on an electroabsorption modulator. The experimental results show that both schemes are useful to mitigate the signal quality degeneration induced by both first and second order PMD. System performance without and with 3R regeneration was separately studied by eye analysis and BER measurements. The signal quality was significantly improved by 3R regeneration under serious 1st-order PMD (up to 40% of the bit interval) combined with 2nd-order PMD (up to ~520ps²). We compared the performance between the two different 3R regenerators. The results show that 3R regenerator using an optical decision gate has a superior performance in synchronously mitigating 1st-order and 2nd-order PMD. Moreover, 3R based on EAM performs wavelength conversion simultaneously, which can be used to avoid wavelength blocking at nodes of the optical networks. Further studies indicate that 3R regenerators have the potential to combat with the effects of PMD combined with PDL and PHB. The limitation of the two schemes is also discussed.

This paper highlights the challenges involved when designing, manufacturing and testing ultra-long high capacity submarine telecommunication systems. We will focus on the transpacific design space and use as an example the design data and measurement results from the most advanced deployed undersea communication link - a 9000 km transpacific segment built using dispersion slope managed fiber. A discussion of recent technical advances and even longer systems, for direct, all optical, wideband connections between the United States and Asia is also included.

An improved scheme for all-optical pulse decision gating by cross-gain modulation(XGM) in semiconductor optical amplifier (SOA) is shown in this paper. Influences of the property of optical clock pulses (including the pulse extinction ratio (ER) and the power) on the shape of regeneration pulse of this optical decision gate are studied by means of numerical simulation and experiment. The power and extinction ratio of optical clock pulses are adjusted in the process of numerical simulation. In the beginning, with the power of clock pulses rising or ER dropping, the pattern of the regeneration pulse gets better and the "1" code noise comes down. There are optimal values of the power and ER of clock pulse for the best pattern of the output pulses. If the ER continues to decrease, the ER of output pulses dropped. Similarly, if the power of clock pulse continues to increase the pattern of the regeneration pulse gets worse. In experiment of optical decision, 40Gb/s regenerated pseudo random bit-sequence (PRBS) data signal was got from degraded PRBS input signal and recovery clock pulses by passing the optical decision gating. The result of experiment is uniform in the numerical simulation.

Intraband crosstalk characteristics of the dilated Benes (DB) network and generally modified dilated Benes (GMDB) network are studied experimentally. The signal-to-noise ratio (SNR) characteristic of the GMDB network is estimated and compared with the DB network. Bit error rate (BER) produced by crosstalk is experimentally investigated and compared. Results show that the GMDB network can eliminate crosstalk influence effectively even though the crosstalk coefficient of the switching element is high.

Several devices for use in high speed fiberoptical transmission systems are discussed. Beside two types of distributed amplifiers, a photoreiver with a tremendous conversion gain bandwidth product is described.

On the following pages we address some actual topics of analog and digital electronic equalisation for 10 and 40 Gbit/s transmission. Analog circuits currently overcome the bandwidth limits for 40Gb/s operation. With first 5 tap feed forward equaliser (FFE) SiGe chips CD and PMD tolerance enhancement can be demonstrated. According to numerical simulations, the combination of the FFE with a one-stage decision feedback equaliser (DFE) should be a good choice to mitigate signal degradation caused by various effects. At 10 Gb/s more complex signal processing based on digitalsignal-processing (DSP) is already implemented in first products as maximum likelihood sequence estimator (MLSE, also referred to as Viterbi equaliser VE) or in a lab prototype of soft error correction (soft FEC). Detailed numerical studies on the performance of the VE reveal that opposite to the analog FFE+DFE, low electrical bandwidth cannot be handled by the standard MLSE scheme and an extension of the processing algorithm is needed. Though the combination of an MLSE based equaliser and soft-FEC has the potential to lead to further improvement, numerical analysis of turbo equalisers combining distortion mitigation and error correction in an iterative way indicate that with a FEC overhead in the range of 7% only moderate equalisation improvement seems to be possible.

We present a theoretical investigation of the nonlinear propagation of an optical pulse in a birefringent photonic crystal fiber (PCF). The strict coupled nonlinear Schroednger equations are solved numerically using a standard split-step Fourier algorithm. The phenomenon of pulse trapping is observed for different polarized angles except 0 degree and 90 degree, when the central wavelength of the input optical pulse is located in the anomalous dispersion region. With a single pulse which inclines from one axis launched into a birefringent PCF, the input pulse is split into two orthogonally components (signal and pump component) between the two orthogonally axes. The signal pulse suffers cross phase modulation by the pump (Raman shifted soliton) pulse and it is trapped and copropagates with the Raman soliton pulse along the fiber. A minimum trapping efficiency is obtained when the polarized angle is at 45^o. For two complementary polarized angles, higher trapping efficiency can be obtained for smaller angle. As the input power of pulse is increased, the red-shift of the Raman soliton is considerably enhanced, leading to further red-shift of the trapped pulse to satisfy the condition of group velocity matching.

GMPLS-controlled photonic network utilizing optical cross-connect in conjunction with WDM equipment will soon be realized in core network to partly eliminate OEO terminations in cut-through traffic. In the next step, all photonic end-to-end connection/switching is emerging technology for realizing higher bit-rate transmission with finer datum granularity, and reducing both size and power consumption of network nodes to establish truly transparent and flexible network. In this presentation, all optical signal processing devices as wavelength converters and regenerators for such future all photonic networks will be reviewed.

40Gbit/s all-optical 3R regeneration was demonstrated. The wavelength of the data signal was converted with an SOA by cross-gain modulation (XGM). The clock recovery was based on Fabry-Perot (F-P) filter of high finesse (F=1000). The actual bit rate of the data signal was 40.04796Gbit/s, matching the free spectral range (FSR) of the F-P filter. Two SOAs were used in series as the decision gate to make full use of the nonlinearity. Taking advantage of the chirp, one narrowband OBF was used to reshape the waveforms in the clock recovery, so as to reduce the pattern effect. Another narrowband OBF was used after the decision gate to diminish the tails of the pulses. The clock signal had a root mean square (RMS) timing jitter of around 843fs. The output data signal of the regenerator had an optical signal-to-noise ratio (OSNR) of more than 40dB/0.1nm. The maximum timing jitter was 1.439ps.

We propose a novel method to compensate the nonlinear distortion from an optical pulse replicator based on an active optical fiber loop. It could be used to faithfully recover the input optical pulse shape from the sampled results of the replicated pulse train, despite of the distortion in the loop. Thus, the replicator could be applied to the measurement of optical pulses for optical fiber communication applications. Our proposed post-processing technique is based on modelling the nonlinear amplification behaviour of the optical amplifier in the loop. The methodology of our technique is described in detail, as well as the simulated recovered results using the proposed algorithm.

We describe some of our recent results on DWDM mm-wave radio-on-fiber (RoF) technology for future broadband wireless systems. A supercontinuum (SC) light source is a promising multiwavelength light source for the system with photonic up-conversion. Multiplexing and demultiplexing schemes for optical frequency interleaving in order to extend the number of antenna base stations can be constructed using a properly designed arrayed-waveguide grating (AWG). We demonstrate a full-duplex WDM mm-wave RoF system using a SC light source. The periodic nature of AWG, where the period is free spectral range (FSR), is used for utilizing the full bandwidth of the SC light source. Half of the SC output modes are used for downlink transmission with photonic upconversion. Another half of the SC output modes are used for uplink transmission with photonic downconversion. These techniques are effective to use a wealth of optical frequency resources from the SC light source. Two-channel downlink and one-channel uplink 60- GHz band RoF signals were simultaneously transmitted over 25-km standard single-mode fiber with error-free and no noticeable power penalty.

The capacity limit of a thermally controlled fiber Bragg grating-based chromatic dispersion compensator, which was initially designed for 10 Gb/s operation, was investigated in a 40 Gb/s system. A CS-RZ DQPSK polarization division multiplex (PolDM) system was used as a testbed. An equivalent quasi error-free 5.94 Tb/s capacity was demonstrated when dispersion of up to 73.8 km of SSMF was compensated. The dispersion slope compensation was satisfactory for C-band operation. Additionally, it was found that the compensator introduced band-pass filtering behaviour, which reduced the compensator bandwidth as the dispersion setting was increased. It was also found that even after 41.5 km, there was around 2 dB penalty introduced to DQPSK system while 5 dB penalty to DQPDK-PolDM, referring to BER of 10^-5.

In this paper, a novel in-band optical spectra and filter shape monitoring technique is experimentally demonstrated. Based on swept coherent detection, the proposed technique simultaneously measures the signal and ASE spectra by adjusting the polarization states of the signal and local oscillator. In our experiment, a high resolution of 0.002 nm is achieved.

The transmission performance in WDM-systems is different for the individual WDM channels. We investigated systematically this variation and wavelength dependence by using semi-analytical simulations. The benefit of this behavior in transparent optical networks is discussed and demonstrated in network case studies.

Optical trains with repetition rates as high as 10GHz~40GHz are obtained by using a novel high speed mode-locked fibre laser. It combines both active mode-locked effect and passive mode-locked effect so the FWHM of the output optical pulses can be decreased to 2ps. And the performance of the proposed fiber laser is analysed.

Numerical simulations have been conducted to investigate the performance of a transmission system in which high spectral-efficiency duobinary modulation format is combined with adaptive electrical equalizer. The conventional filter parameters and the optimum filter parameters are used for comparison of system performance with and without adaptive linear equalizer (ALE). Numerical simulation results show that the dispersion tolerance and nonlinear tolerance (@1 dB EOP penalty) are effectively improved by ALE for duobinary NRZ (DNRZ) with conventional filter parameters, but it is not obvious for DNRZ with optimum filter parameters.

We discuss recent progress on ultra-high-speed modulation formats for next-generation optical transport networks at 160 Gb/s and beyond. In particular, we demonstrate CSRZ, PAP-CSRZ, GAP-CSRZ, RZ-VSB, and DPSK signals at ultra-high data rates.

Group velocity delay ripple (GDR) and reflectivity ripple (RR) are the main nonideal characteristics for chirped Bragg fiber grating (CFBG). The influence of these nonideal characteristics on the CFBG-based transmission systems with carrier suppressed return-to-zero (CSRZ) format was investigated, and the performance of the transmission system was evaluated by the eye opening penalty (EOP) in this paper. In additional, the impact of the phase difference between the central wavelength of the transmitted signal and the GDR on the transmission system which make the transmission performance fluctuate randomly was also studied. The analysis helps to conduct a practical 10Gb/s 2560km error-free transmission experiment in straight line with low power penalty, and no electronics regenerator and forward error correction (FEC) were employed in this system. The experiment result is so attractive in long haul transmission field.

An experimental system used for the chaotic optical communication with the polarization shift-keying (PolSK) modulation technology is presented. In this experimental system, a semiconductor optical amplifier (SOA) based fiber-ring laser is taken as a dynamical chaotic transmitter and the states of polarization (SOPs) of the light wave is modulated through a cross phase modulation (XPM) effect in the SOA by the input message light. The chaotic receiver which is based on a time-delayed configuration can establish the synchronization with the transmitter, so that the transmitted message is restored from the chaotic carrier through a dynamical detection of SOPs of transmission light waves with two polarization detectors. A 5-km long single mode fiber was used as the communication channel. The round-trip time and SOPs of the fiber ring in this system are key parameters to establish the synchronization between the transmitter and receiver. The operation principle of the system was introduced and basic performances of the system were investigated. A chaotic transmission experiment with this system by using a 27-bit RZ sequence at 630 Mb/s was carried out and the message was successfully recovered from the chaotic carriers after 5-km long fiber transmission.

This paper motivates the combination of optical burst switching (OBS) networks with wavelength-routed networks which provide a virtual topology. Opportunities for advanced network design, support for important future transport network scenarios, and resilience as well as capacity adaptation functionality are among its key benefits. The paper discusses principal trade-offs, both qualitatively and quantitatively, regarding virtual topologies and statistical multiplexing in order to derive guidelines for resource efficient topology design. Then, the OBTN architecture is described, which can efficiently transport burst data over a virtual topology of lightpaths and can reduce the port count of optical burst-switched nodes. Finally, the OBTN dimensioning process is outlined and results of a unified performance and cost comparison are presented.

An algorithm to reduce data burst processing delay in group scheduling in core nodes of optical burst switching networks has been proposed. Since, in this algorithm, look-up tables containing all the void time information in scheduling windows are generated as soon as the primary group scheduling session terminates, it becomes faster to reassign dropped data bursts to proper voids, if any, by referring to the tables. The group scheduling with this algorithm showed almost the same performance as the previous group scheduling in regard to channel utilization and wavelength conversion rate. On the other hand, per-burst processing time has been reduced dramatically in the load region of higher than 0.8 and burst loss probability has been improved more under lighter traffic condition.

In this paper, we focus on the study of the characterization of Ethernet traffic and burst assembly algorithm in OBS networks. We investigate the traffic properties before and after burstification in edge router of Ethernet over OBS network. Self-similar Hurst parameter is compared through R/S plot.

Optical Burst Switching (OBS) has been proposed as a most possible switching technology for optical networks. To achieve better performance in the edge node and the whole network, many burst assembly algorithms have been investigated. In this paper, a novel parallel burst assembly mechanism is proposed based on Internet traffic estimation in distributed control wavelength-routed OBS (DWR-OBS), which has a two-way reservation mechanism and can achieve better burst blocking probability while worse end-to-end delay compared with conventional one-way JET OBS. A simulation model is developed to evaluate the performance of the network. Simulation results indicate that with the proposed algorithm network performance improves a lot in DWR-OBS networks.

This paper presents a new approach for quality of service (QoS) support in optical burst switching (OBS) networks. In the approach, the data channels of an outgoing link at a core node are divided into multiple groups, with each group corresponding to a service class. The number of data channels in each group is mainly determined by data traffic. In general, a data burst (DB) can be sent on a data channel reserved by its burst head packet (BHP) only in its own group. Upon failing to reserve any bandwidth in its own group, the BHP tries to re-reserve even preempt bandwidth on a data channel in a lower-priority group. A lower-priority BHP can't reserve bandwidth on any data channel in a higherpriority group. In addition, this paper also investigates the reasonable relation between the preempting DB length and the preempted DB length.

MC-CDMA is a spectrum-efficient modulation in RF systems. Here it is used in optical wireless communications. Performance is simulated by using multiuser detection. Spatial Diversity is also considered and analyzed.

Broadband wireless networks based on a number of new frequency windows at higher microwave and millimeter-wave frequencies have been actively pursued to provide ultra-high bandwidth services over a wireless networks. These networks will have a large number of antenna base-stations with high throughput. Significant reductions in antenna base-station complexity can be achieved if most of the signal routing and switching functions centralized at a central office in the network. In such a network, fiber feed networks can be effectively deployed to provide high bandwidth interconnections between multiple antenna base-stations and the central office. With wavelength division multiplexing, efficient optical fiber feed network architectures could be realised to provide interconnection to a large number of antenna base-stations. In this paper, we present an over view of our recent research into system technologies for fiber wireless networks.

We describe an accurate method to estimate the channel impulse response in diffuse indoor multi-transmitter optical wireless systems. We compare our method with the ceiling bounce method extended for multiple transmitters. Our results show that the channel bandwidth depends not only on the horizontal separation between the transmitter and the receiver, but also on their actual positions.

In free space optical communication, the transmission characteristics through atmospheric channel have important effect on system performance, for example, atmospheric turbulence can cause fluctuations in both the intensity and the phase of the received light signal. The existing system mostly uses intensity modulation and direct detection(IM/DD) because of low complexity and cost, despite atmospheric turbulence can significantly impair link performance and induce a string of error bits. In order to mitigate turbulence effect, coding techniques can be adopted. For a long string of error induced by turbulence units, interlaced code changes the abrupt error into absolute error, thus it can be easily corrected. We analyze the impact of interlaced code on the performance of FSO system whose signal passes through the turbulent atmosphere. Simulations are presented to verify the performance improvement achieved using coding for free space optical communication through atmospheric turbulence channel.

In this paper, we study an architecture of optical burst switching ring network. In our studied OBS ring network, each node is equipped with one tunable transmitter and one fixed receiver (TTFR) for data wavelengths and a pair of fixed transceiver for a control wavelength. Additionally, a new access protocol based on token is proposed. By computer simulations, we evaluate the performance of throughput, queuing delay and probability of transmitting conflicts.

Proposed in this paper is a flexible scheme of all-optical packet header extraction with SOA-based asymmetric Mach-Zehnder interferometer (SOA-MZI). The effect of cross-phase modulation (XPM) and cross-gain modulation (XGM) in SOAs coupled with the interfering characteristics of MZI are employed to make the proposal simple, fast and integrable, with no special treatment of the header pulses. In addition, the scheme, by changing parameters, is applicable to extraction of different header length and bit-rate. Through numerical analysis, an intensity contrast ratio of more than 15dB at 100Gbit/s can be achieved.

Burst contention resolution is one of the most important issues in optical burst switching (OBS) networks. In this paper, an O/E/O conversion and electronic storage mechanism for contention resolution is proposed. Compared with fiber delay lines (FDL) O/E/O strategy can provide much longer and continuous delay time for the contending bursts, which can significantly reduce burst loss rate. For this mechanism, a new burst scheduling algorithm called Shortest Delay- Best Fit (SD-BF) is proposed here to improve bandwidth utilization. In a network, O/E/O can be implemented with FDL to achieve a better performance while reducing node cost. In this paper, a semi-share structure for this combination strategy is proposed to balance the cost and performance. Numerical results show that a better performance is achieved by the combination strategy in the long haul back-bone networks. We also investigate the maximum electronic RAM capacity needed in the nodes to support O/E/O storage, and how to implement Quality-of-Service (QoS) with O/E/O storage.

The most critical challenge in the design of burst optical networks is burst contention, which occurs when multiple bursts contend for the same wavelength on the same link at a certain node simultaneously. The techniques to address this problem include contention resolutions as well as contention avoidance. In contention resolution approaches, conflicts are resolved in wavelength, time or space domain when they have occurred. In contention avoidance approaches, bursty traffic is distributed through routes that are dynamically re-computed or selected according to network congestion status to avoid potential conflicts in advance. Current dynamic routing schemes for OBS are all based on global information, resulting in bandwidth consumption at signaling channels and a considerable delay between network status change and adaptation of the routing tables. In this paper, we propose a novel dynamic routing selection mechanism based on local statistical information, namely dynamic alternative routing, to balance load and avoid contentions in OBS networks. In this mechanism, a route is dynamically selected from the primary path and the alternative route for an incoming burst. The simulation results show the blocking probability of the proposal is significantly reduced and load balancing is efficiently achieved compared to the general OBS.

In this paper, we integrate the pipeline buffering mechanism with the DiffServ-capable OBS to schedule data burst (DB) at core nodes. And we improve a DiffServ-capable virtual scheduling technique (DiffServ-VST) to adapt our scheme. For performance evaluation, we propose a new model to revise the classic Erlang-B formula for calculating burst loss probability in OBS environment: the LAUC-VF-ed M/D/K/K. The simulation results shows that lower packet loss probability and higher link utilization is obtained when traffic load is less than 0.8 by using our scheme. However, the extra end to end delay of AF and BE is brought by pipeline buffer.

Optical Burst Switching (OBS) [1] is now widely considered as an efficient switching technique in building the next generation optical Internet .So it's very important to precisely evaluate the performance of the OBS network model. The performance of the OBS network model is variable in different condition, but the most important thing is that how it works under real traffic load. In the traditional simulation models, uniform traffics are usually generated by simulation software to imitate the data source of the edge node in the OBS network model, and through which the performance of the OBS network is evaluated. Unfortunately, without being simulated by real traffic, the traditional simulation models have several problems and their results are doubtable. To deal with this problem, we present a new simulation model for analysis and performance evaluation of the OBS network, which uses prefetched IP traffic to be data source of the OBS network model. The prefetched IP traffic can be considered as real IP source of the OBS edge node and the OBS network model has the same clock rate with a real OBS system. So it's easy to conclude that this model is closer to the real OBS system than the traditional ones. The simulation results also indicate that this model is more accurate to evaluate the performance of the OBS network system and the results of this model are closer to the actual situation.

Two 2x10 Gbit/s quaternary intensity modulation signals (4-IM) can be generated using quadrature amplitude modulation (QAM), with unequal modulation amplitudes in two orthogonal quadratures. Two 10 Gbit/s NRZ ASK signals and a QPSK modulator allow to generate 4-IM with the same bandwidth as an NRZ-ASK signal. Measured sensitivity at a BER of 10^-9 and chromatic dispersion (CD) tolerance are -21.2 dBm and ~ +130 ps/nm, respectively. Two duobinary 10 Gbit/s data streams and a QPSK modulator allow to generate a 9-point QAM signal, with the same bandwidth as a duobinary signal. A stub filter with a frequency response dip at 5 GHz was used to generate the duobinary signals. Detected as a 4-IM, this scheme features a sensitivity and a CD tolerance of -21 dBm and ~ +140 ps/nm, respectively. A polarization and phase insensitive direct detection receiver with a single photodiode has been used to detect all generated QAM signals as 4-IM signals.

In this paper, the concept of the nonlinear diffusion bandwidth of a fiber is introduced. This simple criterion enables the characterization of nonlinear impairments in single-span transmission systems. Furthermore it is shown how to extend this criterion to multi-span dispersion-managed transmission systems. This enables to easily predict the performance of a given modulation format over various transmission lines with different fiber chromatic dispersion and dispersion maps.

Multilevel intensity modulation such as 4-ASK signalling can reduce the bandwidth requirement in multimode fiber (MMF) links. However, 4-ASK signalling is susceptible to the laser nonlinearity. In addition, even with 4-ASK signalling, the maximum transmission distance over installed MMF is still limited by differential mode delay (DMD) to less than 300m for 10GE. We propose and demonstrate that directly modulated laser nonlinearity and modal dispersion resulting from DMD can be eliminated simultaneously by using a nonlinear electrical equalizer for 4-ASK signalling in high-speed MMF short links.

We present an efficient algorithm for the search of optimum design parameters and transmission quality factor (Q) for a Raman amplified transmission system. By treating the nonlinear phase shift (NPS) as the key parameter for the determination of secondary system parameters, and then scanning the nonlinear Schrödinger equation (NLSE) to get the optimum Q factor as a function of NPS only, we show that the multi-dimensional, multi-parameter, time-consuming design process can be reduced to a highly efficient semi-analytic, 1 dimensional numerical optimization problem. As an application example for the suggested algorithm, we determine the optimum system design parameters (input powers to SMF, DCF, distributed Raman gain, and forward Raman pumping ratio) and Q factor for a single channel 10 G bit/s 2000 km transmission link (SMF-DCF), and then study the effect of pump-relative-intensity-noise (RIN) and span length change to the optimum Q values and changes in the optimum design point. Results show a Q factor improvement for the system more than 1.16dB / 4.89dB at 100km / 200km span length with our design method, when compared to previous optimization method.

Experimental results of a comparative study of two closed-loop systems for suppressing transient gain excursions in an erbium-doped fibre amplifier (EDFA) are presented. One of these systems is referred to as the closed-loop system with the additive pump-control block and the other system is referred to as the closed-loop system with the multiplicative pump-control block. In the experimental results, it is observed that the input pump power needs to be clipped only in the closed-loop system with the additive pump-control block. The pump power clipping is required, as the controller is requesting a negative pump power. In the experimental results, the pump power clipping in the closed-loop system with the additive pump-control block is observed to have a negative side effect in the form of large transient EDFA gain excursions. Thus, the closed-loop system with the multiplicative pump-control block can offer significant advantages over the closed-loop system with the additive pump-control block for suppressing the transient EDFA gain excursions.

A cost-effective tunable dispersion compensator using reconstruction-equivalent-chirp method is fabricated. Only uniform phase mask, sub-micron precision and uniform thin metal film are required in the fabrication. The group delay ripple is less than 14 ps during the whole tuning range. An experiment in 40-Gb/s system is demonstrated with a power penalty of 0.7dB.

Optical burst switching (OBS) is considered as a promising paradigm for the next generation Internet. In the OBS networks, the data channel scheduling algorithm is one of key issues. The object of channel scheduling algorithms is to improve the performance of burst dropping probability and to decrease the complexity of computation. Based on the idea, a new scheduling algorithm, called buffer-based latest available unscheduled channel (BLAUC), is proposed in the paper. Its basic idea can be depicted as follows: to buffer more than one burst control packets (BCP) at first, and then to schedule the corresponding data bursts (DB) by the order of burst arrival time. The max buffer time is an important parameter in the BLAUC algorithm. It is deduced in the paper that the optimisation of the max buffer time is the maximum difference of offset time in the core node. Based on latest available unscheduled channel (LAUC) algorithm, BLAUC only adds buffering and changes the scheduling order. So BLAUC is as simple as LAUC. And adopting the policy of scheduling by burst arrival time, BLAUC has better performance than LAUC in terms of burst dropping probability. Simulation results show that BLAUC not only outperforms LAUC in terms of burst dropping probability, but also outperforms LAUC-VF in the condition of enough buffer time.

Channel scheduling is a key technology in Optical Burst Switch network. In conventional sense, BHP (burst header packet) is always processed at the core node once it is received. Compared with these immediate scheduling algorithms, BHP will be delayed to a time window in Group Channel-Schedule Algorithm before it is scheduled. As for the Group-Channel Schedule Algorithm, analysis was made to investigate the influence caused by the schedule time window ▵t on the performance of OBS network. Then comparison is made between the currently used immediate schedule algorithms and the group-schedule algorithm. Evaluation results show that the packet loss performance is improved.

Optical Burst Switching (OBS) is a promising solution for next generation all-optical WDM networks. It combines the benefits of both optical packet switching and wavelength routing while taking into account the limitation of current all-optical technology. In OBS network, the user data is collected at edge node, sorted based on destination address and Qos requirement, and assembled into viable sized bursts. Prior to transmitting a burst, a control packet is created and immediately sent towards the destination in order to setup an all-optical path for its corresponding data burst (DB). In OBS network, one key factor is implementing switching nodes to achieve low DB loss rate. Another key factor in OBS network is scheduling burst packets on wavelength so as to best utilizing bandwidth. Until now, in most OBS networks, the problem of switching node architecture and scheduling algorithm are considered separately, while they are likely to interfere with each other. At the same time, all existing switching node architecture and scheduling algorithms focus only on resolving contention at local node. Even when the scheduling result may increase contention at some other nodes. In this paper, we propose a scheduling policy that combines both certain switching node architecture and scheduling algorithm. The scheduling policy will not only resolve contention at local node, but will also avoid some contention that may happen at the next-hop node. We call it port modularized scheduling policy (or PM for short).

A novel Mesh optical interconnection board configuration in free-space is considered in this paper. Based on Mesh optical interconnection topology principle, this design is specially designed for integrated chip interconnection on high performance occasion as tiny satellites. The system uses VCSELs as electronic-to-optics converter modules to implement conversion between electronic signals and optical ones. Then optical signals, which have better anti-interference than electronic signals, propagate in free-space in a robust way. And at the interconnected chip side, which could only process electronic signals, optical signals are needed to be converted by PIN arrays into electronic ones. All interconnections are skillfully finished by certain cube prisms and inclined 45^o reflecting mirrors and limited on the optical-waveguide layer inserted in a PCB board. This paper emphasizes on the configuration of an interconnecting unit (considered as a processing element) and its connecting manner. The design aims at light-weight, small-bulk, great anti-interference, advantages at better robustness, lager throughput, especially suitable for data switch, data propagation and IC interconnection on the tiny satellite.

Kanakidis et al presented several kinds of all-optical chaotic communication systems using two encoding techniques and various dispersion compensation maps [1]. It shows that the permitted transmission distances are different for various dispersion compensation maps and various encoding techniques. In order to explore the upper limits of the transmission distance, the parameters of the all-optical chaotic communication system introduced by D. Kanakidis et al. [1] is optimized using genetic algorithm. Some useful results are presented.

Distributed Raman amplification in the transmission fiber is an important technology for advancing the system performance. Co-directionally pumped Raman amplifiers can enhance the performance and allow more flexibility in the system design. However, several sources of non-amplified spontaneous emission noise need to be carefully considered in the amplifier design. The intrinsic cross gain modulation associated with the transient nature of Raman effect can impair the system performance, if the amplifier is not properly designed. We have isolated and measured the impairment due to cross gain modulation in 200-km bi-directionally pumped fiber spans. The penalty depends on the fiber dispersion characteristics and can be small for up to 20 dB on-off co-gain. The benefit of co-directionally pumped Raman amplifiers can be used in multiple long-span transmission to compensate the high loss while maintaining a low nonlinear impairment. It can also be used to extend the length of a single span and achieve a simple system configuration for unrepeatered applications. As opposed to using a span containing multiple fiber types and remotely pumped erbium-doped fiber amplification to achieve transmission over spans with > 60 dB loss, we have demonstrated an unrepeatered link over a single type of fiber with bi-directional Raman pumping. Using the simple conventional non-return-to-zero data format, we achieved transmission of 20 × 10Gb/s channels over a 300-km span of non-zero dispersion-shifted fiber. This simple system configuration provides an important option for terrestrial transmission in remote areas where service access is difficult.

This paper focuses on numerical performance comparison of three typical G.655 transmission fibers in a 40×40Gb/s dense wavelength-division-multiplexing (DWDM) transmission system operating in C-band using return-to-zero differential phase-shift keying (RZ-DPSK) modulation format and hybrid Raman erbium-doped fiber amplifier (EDFA) amplification. An optimization operation over parameters, such as span input power and pre-compensation percent, is performed in the simulation. Simulation results show that the optimum choice of fiber depends on the application requirements and type of the systems and there must be some tradeoffs. Large Effective Area Fiber (LEAF) demonstrates better performance synthetically considering Q-factor, nonlinear tolerance and dispersion tolerance, and all the three G.655 fiber types take on some superiority over the conventional G.652 fiber.

Linear regression method for flat-gain Fiber Raman amplifier (FRA) pump power automatic configuration is proposed. It can give the optimal pump power configuration automatically according to the signal output variation with pump power changes, without any prior knowledge of fiber parameters, like attenuation, effective area, Raman gain factor, and etc. This method is based on the pump power integral model and linear regression theory and is verified by numerical simulation and experiment. The impact on pump configuration caused by power measurement uncertainty is also considered. The results show that deliberate selecting some parameters, like initial pump configuration, pump power variation amount, pump power perturbation times can reduce the impact.

Without using any additional SBS suppression means, 40 GHz 33% RZ pulse train generated by external modulation of a DFB laser diode output was successfully compressed to high-quality 1.3 ps transform-limited soliton pulses using adiabatic soliton compression in 5.45 km long DDF with high SBS threshold. Using the same fiber, as short as 870 fs low-pedestal soliton pulses were produced by higher-order soliton pulse compression.

Ultraviolet communication is a new means of free space optical communication links developed since 1990's. In the paper, we first point out the low data rate drawback of the UV communication system using mercury lamp, and find the limitation arises because of the self trapping of the mercury atomic resonance radiation. Then the working principle and the max modulation speed o f the high data rate UV communication system are explained, at last the high date rate UV communication system which based on the low pressure iodine lamp is described in details. The high UV communication system includes an emitter, a receiver and light propagation channel. The light propagation channel is in the low air. Finally, the system has accomplished good voice and high data rate communication in short distance and its maximal communication rate is 48Kbps which decuple the data rate of the UV communication system made by GTE in 2000. Because of shorter wavelength of the iodine UV source, the whole communication system is more suitable for the security communication usage.

This paper analyse a novel optical multicast scheme without excess optical splitting loss realised using an active vertical coupler based optical switch matrix utilising optical gain in the AVC switch cells. Theoretically analysis and computer simulation are carried out regarding the operational principle and switching characteristics of this scheme. Computer simulation suggests that the amplified spontaneous emission limited scalability of the scheme can be as high as more than 40. Experiment investigation of the predicted multicast switching operation using existing 4x4 switch matrix device has been carried out in for single wavelength and WDM signals. Multicast scale up to 2x4 is demonstrated. Measured switching characteristics and the OSNR values are in good agreement with simulation predictions. High Q factor values for 1-to-4 and 2-to-4 multicasting switched signals indicate that the scale of the multicasting scheme can be further increased.

Exciton saturation and optical nonlinearities caused by phase-space filling(PSF) in InGaAs(P) MQWs are investigated. In-well carrier population under pump beam is calculated as a function of time. Changes in absorption of the two circularly polarized component of the linear polarized test beam resulting from PSF are calculated by the semiconductor nonlinear absorption formula, respectively. The refractive indices of the two circularly polarized component are obtained from the K-K relations. Transient rotation angle of the major axis of the probe polarization is then deduced.

The performance of penalty compensation in 40G bit/s Non-Return-to-Zero (NRZ) systems that use nonlinear adaptive decision-feedback equalizer (ADFE) based on least mean squares (LMS) arithmetic is demonstrated in this paper. There is remarkable performance improvement of dispersion tolerance in single channel 40G bit/s NRZ systems by comparing the 1dB eye-opening penalty (EOP) after the ADFE is introduced. The PMD tolerance limit is improved from 0.17/bit slot of before equalization to 0.22/bit slots, when the system outage probability of tolerant (@BER>10^-9) is 10^-3.

The polarization mode dispersion (PMD) monitoring system is the key integral part of an adaptive PMD compensator. The degree of polarization (DOP) ellipsoid obtained by using a polarization scrambler can be used as either a feedback or feedforward monitoring signal for automatic polarization mode dispersion compensation. Generally, more than several thousands of sampling data of states of polarization (SOP) must be collected to insure getting a correct DOP ellipsoid. This would result in an unacceptable time consuming for adaptive PMD compensation. In this paper, for the first time, we introduce the particle swarm optimization (PSO) algorithm in determining the real-time DOP ellipsoid with high precision, requiring only 100 sampling data of SOPs. Experimental results confirm that the PSO algorithm is effective for ellipsoid data fitting with high precision within 250ms using our hardware environment.

Polarization mode dispersion can decrease the performance of the fiber-optic transmission seriously. Thus the compensation of Polarization mode dispersion is a critical issue in fiber optics. In this paper, a novel Polarization mode dispersion compensator is suggested and demonstrated based on a special-design fiber Bragg grating. A polarization mode dispersion compensation grating with 10-156ps dynamical span and flat-top response is first reported based on the combination of reconstruction algorithm and the equivalent-chirp method. A 10-Gb/s system experiment using the tunable PMD compensator shows the power penalty of BER at is about 1.2dBwhen the PMD of the system is 60ps.

Polarization mode dispersion compensation for fiber communications systems of higher rate is of great importance. Key elements in polarization mode dispersion compensation are polarization controllers and differential group delay lines. Usual ways of compensation for PMD is to control both elements with the same feedback signals, which are processed with an algorithm for control of these elements. Algorithms should be designed for both the controller and delay line with an input signal and give out controlling signals for both elements alternatively. This may cause aimless action of DGD line or polarization state controlling. This paper gives a method of controlling these compensating elements, the polarization controller and the variable differential delay line, with different feedback signals, i.e., to control compensation elements for PMD Compensation separately. It uses Stocks parameters to control polarization controller and uses degree of polarization (DOP) to a variable delay line.

Time Delay Resolution (TDR) of Optical Beam Forming Network (OBFN), is discussed with theoretical calculation and numerical analysis. Time delay ripples apart from ideal time delay exist in single optical true time delay link and delay array based OBFN, but counteract between WDM links. Thus, OBFN is recommended to use WDM architecture.

A novel all optical analog-to-digital conversion system based on fiber nonlinearity was brought forward by Sho-ichiro. In this paper, we analyzed the theory of the system and simulated the process of sampling and quantizing. The split-step Fourier method was used for the numerical simulation. The result of numerical simulation revealed the influence on the sampling and quantizing which was caused by chromatic dispersion and nonlinear effects (XPM, SPM and FWM) in the fiber. Based on the result, the proper parameters of the system were selected.

To improve the all-optical clock recovery, a novel device termed as code mixer was designed to preprocess the injected data signals. In the time domain, the code mixer can change the probability distribution of the amplitude of the data pulses and reduce the amount of codes "0". In the power spectrum, the code mixer can reduce the continuous spectral component. It was demonstrated experimentally that the code mixer could concentrate the probability distribution of the amplitude of the pulses to half of the maximum amplitude. It was also demonstrated theoretically and experimentally that using a code mixer of more ways or several cascade code mixers the clock signal recovered can be further improved. In the experiment of 40GHz all-optical clock recovery using an injection mode-locked fibre ring laser based on a semiconductor optical amplifier, the amplitude fluctuation and timing jitter of the clock pulses recovered caused by the pattern effect were reduced obviously.

The encoding and decoding algorithm of a new type of forward error-correcting code was studied in this paper. With a view to the wider bandwidth and higher signal transmission speed in fiber channel, this kind of forward error-correcting code took more redundant bits as its parity bits and had powerful error correcting function. The encoding and decoding process were simulated on the computer. It was concluded in the paper that the forward error-correcting code with more redundant bits could guarantee the reliability and veracity of the communication system and it could be easily achieved by the software and hardware of an optic modem. The characteristics of this kind forward error-correcting code indicated that it would be widely applied in the military fiber communication system.

Polarization mode dispersion is becoming major system impairment in high speed and long distance optical fiber transmission systems. As the bit rate climbs from 10 to 40 Gb/s per wavelength division multiplexed channel and beyond, optical pulses are increasingly distorted by polarization mode dispersion effect. We report on polarization mode dispersion compensation experiments in 10 Gb/s, 40 Gb/s optical communication systems. The polarization mode dispersion compensator used in the experiments is a compact variable differential group delay element base on concatenation via six magneto-optic polarization rotators (Faraday rotators) of six YVO₄ birefringence crystals whose lengths decrease in a binary power series. Feedback scheme is used to optimize the performance of polarization mode dispersion compensation, using degree of polarization as the feedback signal. In the experiments in 10 Gb/s and 40 Gb/s optical transmission systems, eye-diagrams and bit error rate curves of the code sequences before and after polarization mode dispersion compensation are analyzed. The experimental results demonstrate that the polarization mode dispersion effect induced by the polarization mode dispersion emulator is feasibly mitigated. Separate experiment to reshape the 39ps pulses distorted by polarization mode dispersion is also carried out. The incident optical pulses with width of 39ps are broadened and distorted by polarization mode dispersion effect and then reshaped by the polarization mode dispersion compensator. The relationship between the feedback signal degree of polarization and differential group delay is also analyzed.

We firstly derive the mathematical expression of degree of polarization (DOP) for Gauss pulse sequence, then analysis theoretically the relationship between DOP and differential group delay (DGD). The relevant experiments well demonstrate the relationship of DOP-DGD. We get the conclusions that the DOP-DGD curve is independent of fibre dispersion, line chip, and bit rate, and are affected by the width of optical spectrum and splitting ratio. Based on the theory, we successfully complete an automatic PMD compensation system for first-order PMD, which can compensate PMD in the optical link in real-time. The response time of compensation system can reach several milliseconds.

We propose an adaptive polarization-mode dispersion (PMD) compensation system in which the minor axis of the degree of polarization (DOP) ellipsoid in receiver is used as feedback signal, and the length of the DOP ellipsoid is optimized through the particle swarm optimization (PSO) searching algorithm to compensate PMD. The operation of adaptive equalization is successfully demonstrated by 40 Gb/s transmission system numerically. Furthermore, we show that using DOP ellipsoid as feedback signal and PSO as searching algorithm, the first and higher order PMD can be compensated effectively.

Segment-dividing model is appropriate for research on PMD. Mathematical deduction is made to change it into iterative model. It is easy to find the new model gives clear meaning in physics and reveals the process of PMD accumulation in fiber. It is applicable for being transplanted to analysis on PMD. This iterative method is put forward for practical numeration. When stochastic factor is taken into account on PMD analysis, more elements should be added. Thus statistical mode is put forward. In the statistical mode, the stochastic factor contains two parameters: the mean length of the whole little segments L and the number of the segments N. the choice on the mean length and the number of the segments is a complex problem. Research is done on this problem and results indicate that the choice of L determines the veracity of the statistical model.

Effective utilization of highly nonlinear fiber (HNLF) with relative large normal dispersion to further broaden optical spectrum of 40-GHz, 1.2-ps optical pulse train is investigated, in which an optimized scheme is presented. In the scheme, the pulse is launched into the HNLF of the first section to generate the spectrum broadened linearly up-chirped pulse, and recompressed after propagating through a suitable length of conventional single-mode fiber (SMF) of the second section for the pulse chirp being compensated. At last, the pulse travels through the second HNLF of the third section for further optical spectrum broadening. This new scheme applies the chirp pulse compensation technique and has scalability to the peak power of seed pulse. The experimental results show that ~1.7 nm increment of 3dB spectrum bandwidth is obtained when the new scheme is implemented given the same HNLF length and input seed pluses, and the coherence properties of the original pulses are maintained. This scheme provides a more effective means of pulse spectrum broadening in the HNLF when the peak power of seed pulse is confined by the EDFA output power and the high repetition rate of pulse laser.

The optical multilevel modulation is one of the attractive candidates to significantly increase the channel bit rate and total capacity of future optical fiber communications. We review various multilevel modulation schemes proposed to date, including our experimental results of the 8-level and 16-level APSK modulation/demodulation schemes at 10 Gsymbol/s. We also discuss the advantages and disadvantages, and future issues of the optical multilevel modulation schemes.

A comprehensive study of the noise characteristic in optical balanced receivers in differential phase shift keying (DPSK) systems is performed. The receiver model based on Karhunen-Loève (KL) transform which may include the main transmission impairments in WDM systems is proposed as well as the numerical computation techniques for bit error ratio (BER) assessment. The statistics of the decision current is obtained which shows exact agreement with that of Monte-Carlo simulation, so the analytical result is reliable for accurate estimation of BER. After validated by the time consuming Monte-Carlo error count, the proposed model is adopted in a back-to-back simulation and the 3 dB advantage of DPSK over traditional on-off keying (OOK) is clearly verified.

All-optical 40Gbit/s NRZ-to-RZ format conversion scheme with broad signal conversion bandwidth is presented based on the cascaded second-harmonic generation and difference-frequency generation (SHG+DFG) in a periodically poled lithium niobate (PPLN) waveguide. The considered device is a Mach-Zehnder interferometer (MZI). The conversion mechanism relies on the combination of amplification and nonlinear phase shift induced on the signal field during SHG+DFG processes, which making the MZI get unbalanced. The conversion process is numerically calculated with a good performance achieved when ignoring group-velocity mismatching, showing that NRZ-to-RZ format conversion can be successfully realized with wavelength unchanged. The nonlinear phase shift induced on the signal field is analyzed in detail. It is found that large values for nonlinear phase shift can be achieved when quasi-phase-matching (QPM) for SHG but appropriately phase mismatched for the DFG process. Finally, the effect of the nonlinear phase shift induced on the signal field on broadening the signal conversion bandwidth is further discussed. Approximately, 94nm conversion bandwidth can be achieved by keeping pump-wavelength unchanged while tuning signal wavelength, which provides the possibility of realizing a broadband tunable NRZ-to-RZ format conversion.

This paper studies advanced optical phase modulation formats in 40Gb/s Ultra-long-haul systems. As well known, the performance of 40Gb/s Ultra-long-haul systems depends upon the modulation formats. Since DPSK modulation format has higher spectral efficiency and more tolerance to fibre nonlinearity induced impairment, different duty cycle has important impact on result. In this letter, we comprehensively analyzed the transmitting performances of optical phase modulation formats with using the eye-opening penalty (EOP). NRZ-DPSK, full frequency modulated RZ-DPSK (FullRZ-DPSK), half frequency modulated RZ-DPSK (HalfRZ-DPSK) and CSRZ-DPSK modulation formats was numerical simulated within four kinds fiber system: G.652 Fiber, True Wave fiber (TW), True Wave-Reduced Slope fiber (TW-RS) and Large Effective Area Fiber (LEAF). Through modelling and simulation, we compute the EOP of these phase modulation formats, with different average optical input power. The numerical simulation result shows thatCSRZ-DPSK is best performance in all phase modulation, and G.652 outperforms other types of fibers in 40Gb/s Ultra-long-haul systems.

A new class of multilength, constant-weight two-dimensional multiwavelength optical orthogonal code (2D MWOOCs) with large capacity and good correlation properties is constructed based on multilength one-dimensional (1D) OOCs. The performance of these new MWOOCs in an optical code division multiple access (OCDMA) network with double-services is analyzed. The result shows that media with the shorter codewords performs much better than the media with longer codeword, and OCDMA system with these new multilength MWOOCs performs well. These features allow large capacity multimedia transmission in OCDMA system.

This paper investigates the survivable traffic grooming problem for optical mesh networks employing wavelength-division multiplexing (WDM). While the transmission rate of a wavelength channel is high, the bandwidth requirement of a typical connection request can vary from the full wavelength capacity down to subwavelength. To efficiently utilize network resources, subwavelength-granularity connections can be groomed onto direct optical transmission channels, or lightpaths. Meanwhile, the failure of a network element can cause the failure of several lightpaths, thereby leading to large data and revenue loss. Fault-management schemes such as protection are essential to survive such failures. Different low-speed connections may request different bandwidth granularities as well as different protection schemes. How to efficiently groom such low-speed connections while satisfying their protection requirements is the main focus of our investigation. The paper tackles the dynamic survivable traffic grooming problems in multifiber wavelength-routed optical networks by representing the network as a layered graph model. This graph multi layers, where each layer represents a specific wavelength. Each link in the layered graph has more than one fibers and an associated cost. We use a modified Dijkstra algorithm that has a reduced complexity due to the structure of the layered graph. Heuristic algorithms for fiber selection based on a well-designed link-cost metrics are proposed. The performance of various routing algorithms is evaluated through simulation studies.

Optical layer capacity and unit cost improvements are basic to the rapid growth of IP networks. However, the new rapid reconfiguration and restoration capabilities of the optical layer have been sparingly utilized by IP network operators. This is consistent with the economics: the widely deployed optical layer architecture based on a "discrete" optical crossconnect (DOXC - one not integrated into the WDM) incurs heavy interface costs. In addition, there are architectural and control issues which are roadblocks to IP exploitation of rapid optical layer agility. In the first part of the paper, we describe a next generation all-optical architecture based on OXCs integrated with the WDM ("IOXCs"), and one instantiation of this architecture using a class of reconfigurable degree-N optical add-drop multiplexer (OADM). A new shared mesh restoration mechanism based on hot standbys is designed to overcome slow convergence and transient behaviors that are typical in all-optical networks. A series of economic comparisons are made on both a 120-node hypothetical national network and a smaller express backbone network to demonstrate the cost effectiveness of the new IOXC-based restoration in comparison to the IP layer restoration after some optical layer failure that results in one or more IP link failures. In the IP over Optical architecture, the Optical Layer can provide its IP clients rapid and efficient restoration for IP link failures that are due to some optical layer failure; however its inability to protect against router and router interface failures erodes its attractiveness. In the second part of the paper, we propose a joint IP/Optical restoration mechanism to deal with this. It is suitable for IP offices with at least two backbone routers and an optical cross-connect. The OXC is used to re-terminate OC-48/OC-192 links from a failed backbone router to its mate. The entire interoffice portion of the re-terminated link is reused, thus minimizing the amount of additional inter-office capacity required to protect against router failures. The proposed scheme is also applicable to restoration for router interface failures as well as to cases with single backbone router office architecture.

We demonstrate a simple method of generating time division multiplexed pulse trains with correlated and controlled phase, only using low-cost birefringent crystals and polarizers. As an illustration of this method, we achieve carriersuppressed RZ (CS-RZ) pulse trains at 20GHz and 40GHz respectively by multiplexing 10GHz ultrashort pulse train. Furthermore, by numerical simulations, it is verified that the suitable phase-modulated pulse train generated by our method can be applied to high speed optical transmission system for suppressing intrachannel nonlinear effects.

A novel all-optical format conversion scheme from NRZ to RZ based on sum-frequency generation (SFG) in a periodically poled LiNbO₃ (PPLN) waveguide is proposed, using a nonlinear optical loop mirror (NOLM). The conversion mechanism relies on the combination of attenuation and nonlinear phase shift induced on the clockwise signal field during the SFG process. The SFG between pump, and co- and counter- propagating signals in the PPLN waveguide are numerically studied, showing that counter-propagating SFG can be ignored when quasi-phase matching (QPM) for SFG during co-propagating interaction. The nonlinear phase shift induced on the clockwise signal field is analyzed in detail, showing that it is more effective to yield large values for nonlinear phase shift when appropriately phase mismatched for the SFG process. Two tuning schemes are proposed depend on whether the sum-frequency wavelength is variable or fixed. It is found that the latter has a rather wide 3dB signal conversion bandwidth approximately 154nm. Finally, the influence of reversible process of SFG is discussed and the optimum arrangement of pump and signal peak powers is theoretically demonstrated. The result shows that proper power arrangement, pump width, and waveguide length are necessary for achieving a good conversion effect.

A 7-chip, 280-Gchip/s quaternary phase-shift code is assigned to short pulses after reflection from a encoder, The code is then recognized by a decoder. The encoder and decoder used within our experiments are 4-phase shift SSFBGs. Each grating containing 7-chip has a uniform amplitude refractive index level along its length but in which discrete jumps in phase (0,∏/2,∏,or 3∏/2) are written into the grating at the boundaries of adjacent spatial chips. The 4-phase shift SSFBGs were fabricated using "equivalent phase shift" method, wherein the desired phase shift in one channel of Sampled Bragg grating (SBG) could be achieved by simply changing the sampling period. Only ordinary phase masks and submicrometer precision are needed to fabricate the encoding and decoding SSFBGs. The quaternary phase encoding and decoding of short pulses at 40Gb/s is experimentally demonstrated, to the authors' knowledge, this is the highest single channel rate obtained from FBG based OCDM system. Multiple user interference (MUI) is also considered and evaluated, we present an experimental demonstration 2-user 40Gb/s/user (2×40Gb/s) OCDM system employing 4-phase pulse pattern generation and recognition. In addition, a nonlinear optical loop mirror (NOLM) within the receiver is introduced to improve the system performance under multiuser operation, we show that the NOLM can act as a nonlinear processing element capable of reducing both the pedestal associated with conventional matched filtering and the width of the associated code recognition pulse. The system benefits of using the NOLM are experimentally demonstrated under 2-user operation.

Chromatic dispersion (CD) in single-mode optical fiber distorts pulses and is a big obstacle against the upgrading of long-haul, dynamically routed wavelength division multiplexed (WDM) transmission systems at 10 Gbit/s and beyond. High-performance adaptive dispersion compensators are required as well as reliable low-cost hardware for the detection of residual CD. Targeting high-capacity metro systems, full-band 1.6-Tb/s (40x40Gbit/s) adaptive CD compensation is demonstrated in this experiment, using CSRZ-ASK and NRZ-DPSK modulation formats. A multichannel dispersion compensator, tunable in the range -700 to -1500 ps/nm, is automatically controlled by arrival time detection in one of the 40 transmitted WDM channels. Dispersion and its slope are tuned simultaneously by a thermal gradient of the grating-based compensator to match the parameters of standard single-mode fiber (SSMF) with lengths between 44 and 94 km.

The CSO (composite second-order distortion) induced by self-phase modulation (SPM) in an AM-VSB externally modulated CATV system which employs optical fiber Raman amplifiers (FRA) is studied theoretically in this paper. Firstly, the analysis of CSO is achieved by solving Schrödinger equation using perturbation method. And then numerical results are presented and discussed. According to the study in the paper, the SPM induced CSO performance can be improved by more than 10 dB by using FRA than using EDFA in CATV systems.

Cost-effective short-haul but high-capacity optical transport systems are becoming increasingly important for metropolitan-area and access applications; examples are metro-feeders, inter- or intra-office links between routers and cross-connects, and storage-area networks. Transmission links of this kind are characterized by distances of 10 to 100 km and typically use directly modulated lasers providing a high output power at a low device cost. Due to the inherent wavelength drift of uncooled system components, full spectrum CWDM is defined for up to 16 or 18 channels on a coarse 20-nm wavelength grid between 1270 and 1610 nm. To meet the demand for the increase of transmission capacity in the metro and access domain, there are several potential strategies to remove the capacity limit of 16×2.5 Gb/s found in conventional CWDM systems while still retaining the cost-effectiveness and the modular "pay as you grow" philosophy of CWDM. In this paper, we provide an overview of recent progress in the field of CWDM upgrade techniques. CWDM architectures with increased per channel bit-rates and sub-band DWDM channel overlays are reviewed and experimental results are presented. The particular combination of CWDM with DWDM sub-bands and bit-rate upgrades is expected to deliver the highest capacity gain. However, we also show that the suitability of the system for higher bit-rates hinges on the following considerations: the availability of low-cost 10-Gb/s DMLs at the respective wavelengths, the link budget and the chromatic dispersion of the deployed system.

Differential phase-shift keying (DPSK) has recently attracted a great deal of attention especially for high-speed, long-haul transmission systems. The reason is that, compared to conventional on-off keying (OOK), this format offers ~3 dB better receiver sensitivity and higher tolerance to some fiber nonlinearities. To take full advantage of this modulation format, however, balanced reception should be provided for the receiver, which makes the system implementation complex and costly. This paper reviews the recent progress of DPSK format, with emphasis on system impairments by linear and nonlinear optical phenomena. We also compare the merits and demerits of DPSK with those of OOK format.

We review physical layer challenges associated with the emerging class of optically-routed networks. We discuss the influence of reconfigurable optical add-drop multiplexers (ROADMs), advanced modulation formats, and electronic signal processing techniques at transmitters and receivers to increase optical transparency in optically-routed networks.

We experimentally demonstrate optical packet switching with less than 8 ns switching speed for a 10 Gb/s RZ payload and a 155 Mb/s NRZ label using a 4×4 optical crosspoint switch (OXS) matrix. The wavelength dependence of the OXS is also investigated in terms of the receiver sensitivity and the output OSNR. Very good switching performance can be achieved from 1535 nm to 1561 nm. Our results suggest that the OXS can be a promising device for switching future high speed impulse coded signals.

We present system-level characterization results of a MEMS tunable optical filter with a flat-top passband and narrow transition bands. The proposed optical bandpass filter can continuously change its optical bandwidth and center wavelength using a free-space grating and a variable-aperture MEMS reflector. We observed that the receiver sensitivity degradation is within 1 dB across the 6-dB optical passband.

We study the design issues of a node in multi-granularity all-optic network based on a proposed dynamic multigranularity including coarse granularity (i.e., fiber and band) and fine granularity (i.e., wavelength) traffic model. An optimization design plan is proposed by analyzing the associated parameters of MG-OXC architecture in detail and optimizing the configuration of them. The proposed optimization design rules can significantly reduce the node complexity and improve the blocking performance of arriving requests, which lead to smaller device size and lower cost in all-optical networks.

An all-fiber magneto-optic switch is designed in this paper, which makes use of Faraday Effect, fiber-type polarizing beam splitter/combiner (PBS/PBC), magneto-optic crystal fiber, nanosecond impulser and high-speed magnetic field control technology. The design scheme uses magneto-optic crystal fiber instead of bulky magneto-optic crystal. The optical route design includes linear layout of magneto-optic crystal fiber, the optical route design of polarized light in fiber PBS/PBC, magnetic route design and analysis of linear solenoid, etc. A solenoid which is driven by nanosecond current and can generate high-speed magnetic field is designed and manufactured. The result analysis shows that pulse signal can be greatly strengthened by impressed-bias static magnetic field which is perpendicular to transmission direction of light beam. Static magnetic field insures the intensity of induced magnetization where the light passes come back to original value, so that it can offer high-extent magnetization equality. The intensity of magnetic field can be improved to two orders when the length ratio of magneto-optic crystal fiber to solenoid is increased eight times. When the diameter of magneto-optic crystal and solenoid is decreased from three millimeters to one millimeter, the intensity of magnetic field can be improved to one order. The all fiber magneto-optic switch is slim in structure, flexible, easy to high-density integration and expansion, so there are many integration methods to design all-fiber magneto-optic switch array, which is characterized of smaller bulk, higher magnetic field usage, lower consumption and driving voltage, etc.

Recently, FTTH has finally emerged from the R&D stage and became a commercial reality. For example, in Japan, it has been reported that the total number of FTTH subscribers exceeded 1.5 million in 2004. In US, there have been numerous announcements of FTTH deployments by both operating companies and municipalities. It appears that this trend is rapidly spreading all over the world. At present, most of the FTTH deployments utilize TDM PON (such as GPON and EPON). However, WDM PON is beginning to attract significant attention, as Korea started a large-scale field traial this year. In fact, WDM PON has long been considered as an ultimate solution for the access networks due to its large capacity, easy management, network security, and upgradeability. In this paper, we review the current issues in WDM PON and report the relevant technical progresses achieved at KAIST. The subjects to be covered include low-cost WDM light sources (such as spectrum-sliced incoherent light sources and ASE-injected Fabry-Perot lasers), techniques for the delivery of broadcast services in WDM PON, wavelength-tracking technique for the wavelength-selective devices placed at the un-powered remote node, survivable WDM PON architectures, and fault monitoring and localization techniques.

A new code structure for spectral amplitude coding optical code division multiple access system based on Double Weight (DW) code families is proposed. The DW has a fixed weight of two. Enhancement Double Weight (EDW) possesses ideal cross-correlation properties and weight can be any odd number which is grater than one. It has been observed that theoretical analysis for EDW code is much better performance compared to Hadamard, and Modified Frequency Hopping (MFH) codes.

The coherent OCDMA system could suffer from severe multiple access interference (MAI) and beat noise, which limit the maximum number of active users that can be supported in a network. One effective method to reduce the beat noise as well as the MAI noise is to lower the interference level by adopting ultra-long optical code. Applying optical thresholding technique is also crucial to enable data-rate detection for achieving a practical OCDMA system. In this paper, we review the recent progress in the key enabling techniques for asynchronous coherent OCDMA: the novel encoder/decoders including spatial lightwave phase modulator, micro-ring resonator for spectral phase coding and superstructured FBG (SSFBG) and AWG type encode/decoder for time-spreading coding; optical thresholding techniques with PPLN and nonlinearity in fiber. The FEC has also been applied in OCDMA system recently. With 511-chip SSFBG and SC-based optical thresholder, 10-user, truly-asynchronous gigabit OCDMA transmission has been successfully achieved. Most recently, a record throughput 12×10.71 Gbps truly-asynchronous OCDMA has been demonstrated by using the 16×16 ports AWG-type encoder/decoder and FEC transmit ITU-T G.709 OTN frames.

We demonstrate that all the inter- and intra-channel nonlinear impairments can be eliminated simultaneously by optical phase conjugation (OPC) in a power-symmetry system. However, for practical systems without power-symmetry, it is found that the effects of OPC on various nonlinearities are different in the same link. Even some nonlinearities are suppressed and some are enhanced. Therefore, optimizing the transmission link with OPC to suppress the dominant nonlinearity is demanded. By using 1-km-long highly nonlinear fiber (HNLF) and tuning the pump wavelength near the zero dispersion wavelength of the HNLF, we experimentally generate the phase conjugation of the dispersed ~300 fs pulses. OPC with conversion efficiency of about -16 dB and conversion bandwidth of about 38 nm is obtained.

Optical wireless communication (OWC) is beginning to emerge in the telecommunications market as a strategy to meet last-mile demand owing to its unique combination of features. Turbo codes have an impressive near Shannon-limit error correcting performance. Twofold turbo codes have been recently introduced as the least complex member of the multifold turbo code family. In this paper, at first, we present the mathematical model of signal and optical wireless channel with fading and bit error rate model with scintillation, then we provide a new turbo code method to use in OWC system, we can obtain a better BER curse of OWC system with twofold turbo code than with common turbo code.

Here we study the design rule for optical links in the metro arena. We discuss the basic power paradigm, the related physical effects, the improved system testing methods, and the system margin. The test results for the SFP transceiver are also reported.

Next-generation wavelength routing optical networks requiring optical cross connects (OXC) in the network have the ability to direct optical signals from any input interface to suitable output interfaces by configuring their internal embedded optical switch matrices. Optical switches based on MEMS technology have the fundamental advantage of being able to exploit the benefits of free-space interconnection (including low loss and crosstalk, and low polarization- and wavelength-dependence), together with the advantages of integrated optics (including compactness, optical pre-alignment, and low cost ). Therefore, it offers the possibility of achieving high port-count in a small, low-cost system with excellent optical quality. In this paper, at first the basic concepts of the two kinds of optical MEMS switches--2-D and 3-D switches are reviewed and the principle of the MEMS switches is introduced. The key parameter in the assembly/packaging of optical components is the insertion losses due to misalignment between the optical components. The misalignment losses are calculated as the coupling coefficient between two Gaussian beams having certain lateral and longitudinal separations and a tilt angle. It's concluded that optical design of a free-space MEMS component is a compromise between lateral misalignment and angular misalignment from the simulations. At last the application of optical MEMS switches in the optical cross-connects is discussed including the configurable WADM (wavelength-add-drop multiplexer) which can be achieved using 2-D MEMS crossbar matrix switches and the three-dimensional (3-D) microoptical switching system (3D-MOSS).

A reconfigurable optical add-drop multiplexer (OADM) based on F-P cavity and piezoelectric translator is proposed. In this structure, the displacement of F-P cavity can be controlled by piezoelectric translator. The OADM can add/drop one of the multi input channels or pass the channel directly by adjusting the electric voltage of the piezoelectric translator. The channel isolation is more than 30dB. Crosstalk, including inter-band crosstalk and intra-band crosstalk, in F-P filter based OADM are analyzed and calculated in this paper.

A novel type of optical spreading sequences, named the 'new-Modified Prime Code (nMPC)', is proposed for use in synchronous direct-detection optical code-division multiple-access (CDMA) systems which employ both pulse position modulation (PPM) and overlapping pulse position modulation (OPPM) schemes. The upper bounds on the bit error rate (BER) for nMPC used in PPM-CDMA systems are derived and compared with the respective systems, using a modified prime code (MPC) and a padded modified prime code (PMPC). The nMPC is further applied to the OPPM-CDMA system and the system with a proposed interference cancellation scheme. Our results show that under the same conditions the PPM-CDMA system performances are more improved with the use of nMPC than with the two other traditional codes. Moreover, they show that the system performances are significantly enhanced by the proposed interference reduction methods, if the nMPC is used in the OPPM-CDMA systems.

Finding the right technology to cover the last few miles of any network has always been a problematic challenge for service providers. Whether wireless, copper or optical links are to be used, there's inevitably a set of competing technologies to choose from. Finding the optimum solution can be a complex process, with numerous interlinked factors to be taken into account. In this context, developments underway in the world's market around the use of high speed Passive Optical Networks (PONs) are going to be particularly significant - especially as far as the role of Gigabit PON (GPON) systems are concerned. The newly standardized ITU-T G.984 GPON technology, is delivering extremely high bit rates while still supporting the transmission of native formats such as IP and TDM at extremely high levels of efficiency. Due to its unprecedented offered bandwidth, GPON is the ideal technology for large-scale FTTH applications where multiple end-users are requiring an ever-growing bandwidth. Moreover, in areas populated by both business and residential customers, GPON is the most cost-effective solution.

Interferometric noise(IN) is essential to an OCDMA system, since the tree topology of optical coder and decoder induces the multi-path transmission of signal from same laser. Signal noise ratio(SNR) including MAI, optical beat noise(OBN) and IN is firstly deduced in this paper. Numerical and experimental results show that in one dimension OCDMA system with narrow line width laser, interferometric noise is great enough to limit system performance with relative high bite rate and small frequencies' difference; while in tow dimension OCDMA system, interferometric noise is a minor factor compared to OBN.

Since the deployment of optical communications networks, carriers have used many mechanisms to protect fiber in the core, because of the large volume of traffic carried in the core, and the affect of a failure on the entire user community. Access links, connecting end users to the core network were not protected because they affected a small number of customers, carried relatively low bandwidth traffic and were less critical. As enterprises use ever growing bandwidth and become increasingly dependent on their network connectivity, they demand protected services. Since fiber breaks can occur anywhere in the network, including the so-called "last mile" carriers are looking today for protection systems for the access and edge network links. This paper focuses on the emerging solutions for this application, detailing the key criteria for selecting fiber protection switches: optical performance and cost effectiveness (important because the cost must be amortized over a small number of customers). It discusses solutions to the various access network topologies and common protection schemes.

An urban wireless optical SAC-CDMA network with MQC is proposed and analyzed. Analytical techniques for performance valuation of synchronous random access packet switching in urban wireless CDMA system are presented. Considering the nodal architecture design at higher layers, as well as the characteristics of the physical layer, such as atmospheric scintillation, phase-induced intensity noise and receiver noise, steady-state throughput characteristics using Binomial Arrival Model are obtained. Numerical results indicate that use of multi access coding can provide utilization-delay characteristics superior to that of ALOHA, and the atmospheric scintillation is an important factor affecting the throughput of urban wireless optical CDMA network when the atmospheric turbulence is comparatively strong.

A modified carrier suppressed return-to-zero (CSRZ) format was forward in this paper, and it helps to improve the extinct ration (ER) and signal-to-noise ration (SNR) of the transmitted signal. Experiment of 2560km error-free transmission in straight line without forward error correction (FEC) and electric regeneration was demonstrated in this paper to evaluate the performance of modified CSRZ, and chirped fiber Bragg grating (CFBG) employed as the dispersion compensator in this transmission system. The experiment result showed that if the modified CSRZ format was applied in this 2560km transmission, less than 2.5dB power penalty was cost after 2560km transmission, and on the other hand, error-free transmission could not be got in the conventional NRZ-based CFBG-compensated system.

In this paper, we introduce a novel method for free-space optic communication. Normally, the free-space optic communication is sensitive to atmospheric turbulence. In order to overcome the weakness, we utilize two differential modulation optical beams, which improve markedly the quality of free-space optic communication. The experiments, based on two wave mixing scheme were¹ done using photorefractive crystal LiNbO3. Photorefractive grating will be changed with the electrical signals, which will results in the energy transfer process between two beams, and then form two differential modulation optical beams. Firstly, we will present a brief introduction to the theoretical arguments for energy transfer process in two wave mixing experiments. After briefly explaining the method to achieve two differential modulation optical beams we discuss the experiments for free-space optic communication. Finally, we conclude that the novel method improve markedly the quality of free-space optic communication.

The implementing of optical virtual private networks (OVPNs) over an optical burst-switching (OBS) network based on MPLS is discussed. The network architecture, the core and access nodes faculty are described. We also propose a new model with three traffic priorities, which used to analysis the delay and blocking performance of OVPN over OBS/MPLS network based on queuing theory.

The performance of long-haul dispersion managed DWDM soliton systems is systematically studied. It is found that the values of Q factor and optimal window for some channels are very different. The optimized bandwidth of inserted optical filter and dispersion map are suggested.

A scheme of WDM+OCDMA system employing optical hard-limiter(OHL) is proposed, and its performance of normalized throughput is analyzed. Upper bound of normalized throughput is obtained when all simultaneous users are equally allocated to different wavelength channels. Lower bound of normalized throughput is obtained when all simultaneous users are firstly allocated to same wavelength channels. Compared with the performance of WDM+OCDMA system without OHL, both upper bound and lower bound of normalized throughput in WDM+OCDMA system with OHL can be improved.

It is essential to increase or decrease storage time of a Dual Loop Optical Buffer (DLOB) by properly selecting fiber loop length. We experimentally investigate the performance constraint of DLOB owing to the Four-wave mixing (FWM). Our results show that some parameters, such as the control signal wavelength, length of fibre loop, types of fibre have strong impacts on the performance of DLOB. The storage time can be increased by power compensation, but the signal become degraded after several loops due to FWM. To decrease the buffer time, the power of the control signal is increased and high-nonlinear fibre is selected to shorten the fibre loop, but the FWM can not be eliminated and has a critical impact on the performance of DLOB.

Since IP-over-WDM has become a natural choice for the next generation backbone networks. It is more eminent to implement efficient multicast in WDM networks. A light-tree has been defined to support multicast sessions in WDM layer, it construct a point-to-multipoint light-paths from a source to multiple destinations and can be implemented by power-splitters. With the mature of power-splitters, there is a need to develop framework at higher layers to efficiently use the multicast available at WDM layer. In this paper, we concentrate on OBS-based multicast, new node models supporting WDM multicast are proposed. Then base on the models, we put forward novel implementation architecture of GMPLS-based OBS network to support multicast services, which maps an IP multicast-tree onto a multicast LSP that are connected to form a WDM light-tree. Two salient features of the architecture are: multicast LSP is constructed by data-driven; and the inter-operation between GMPLS framework and OBS. Finally, we discuss the performance of the proposed architecture.

Secure optical communication systems based on general bent sequences have been described which encode/decode information with arrays of fiber Bragg gratings (FBG). A broadband pulse coming from an optical source will be split into a series of separated pulses when it reaches the encoding FBG array. These pulses have different spectrums and different time delays. Because it is difficult for an attacker to recover the original pulse if he doesn't know the exact structure of the encoding FBG array, one can say that the information from the optical source has been encrypted. The coded signal will be recovered by the decoding FBG array having the inverse structure of the encoding FBG array. Because various structures of encoding FBG arrays could generate different reflected pulses to ensure the security of communication in the unsafe channels, grating tuning techniques are used broadly to increase the modes of the systems. Some models of the systems will be introduced and their security will be discussed. Bent sequence is generated by a general class of bent functions to charge the modes of FBG arrays in order to increase the security of our systems because of its excellent characteristics in cryptology. That makes the optical communication system safe against the interception. Simulation and discussion will be included and the applicability of the system in the future has been prospected.

In this paper we compare differential quadrature phase-shift-keying (DQPSK) and quaternary differential phase amplitude-shift-keying (QDPASK) transmissions in 80Gb/s optical communication systems with highly nonlinear fiber (HNLF) channel and evaluate their tolerance against channel nonlinearity and receiver interferometer phase error (IPE). The results show that QDPASK signal has better receiver IPE tolerance while DQPSK signal has better channel nonlinearity tolerance. The performance of DQPSK system is better than QDPASK system only when IPE is not serious.

We propose a parallel interference cancellation technique for two-dimensional optical code division multiple access (OCDMA) system. In the proposed system, we estimate the multiple access interference (MAI) by group information codes. This new interference cancellation technique can be realized with simple structures. Performance analysis shows that it can improve the bit error rate (BER) performance, and increases the number of simultaneous users in the system.

A novel all-optical header extraction scheme based on cross gain modulation and self-frequency shift in SOA is proposed and demonstrated in this paper for the first time. In the proposed technique, the header is assumed to be encoded in 2.5Gb/s PRBS and inserted ahead of a payload with 40Gb/s ratio. With theoretical simulation, the feasibility and property of the proposed scheme are demonstrated and analyzed respectively. Results indicate the performance of the technique is influenced greatly by the parameters of SOA and optical bandpass filter (OBF), such as the carrier lifetime and the 3-dB bandwidth etc. With properly chosen parameters of SOA and OBF, the transmitted energy of the payload can be suppressed effectively and the header can be extracted successfully from the input packet.

The performance of normalized throughput in WDM+OCDMA is compared with that of MW OCDMA system after considering the effect of chromatic dispersion. Normalized throughput in MW OCDMA system will be reduced after considering the effect of chromatic dispersion. Both WDM+OCDMA and MW OCDMA systems have comparable performance of normalized throughput when traffic load is light. However, WDM+OCDMA system has better performance when traffic load is heavy. Furthermore, the peaked normalized throughput of WDM+OCDMA system is larger than that of MW OCDMA system.

Traditional burst assembly method only differ service class priority by increasing extra offset time so as to assure QoS at Optical Burst Switching (OBS) core node which increases all network delay extremely, especially for optical network with many hops and large radius. In order to take full advantage of flexible buffer and control function of electronics device and optical large capacity predominance, it is desired to process real-time traffic at edge ingress node and guarantee delay fairness under small enough delay and QoS based on offset time differentiation for various classes. To achieve lower assembly delay at OBS edge node and offset time based QoS guarantee at core node, this paper proposes an approach called Prediction and Offset QoS Assembly (POQA) to improve existing burst assembly mechanism at edge node. POQA method takes into account the delay fairness and QoS-based offset time adequately. The ratios of assembly time and offset time are introduced to make a delay and offset differentiation among various traffic classes. Assembly delay, burst size, burst utilization and length error of various classes are compared under different offered load. Moreover, in contrast with traditional time assembly method, simulation results show that POQA method can achieve a significant improvement in terms of burst delay and utilization and realize the delay fairness and offset time QoS differentiation under different traffic offered load.

The dispersion managed soliton transmission is a new soliton transmission scheme, but the ultra long-haul WDM system is limited its communication capability for multi-perturbations. In the present, the multi-perturbations including the amplificatory spontaneous emission (ASE) noise, across phase modulation (XPM) between the channels etc are considered. The evolvement characteristics of dispersion managed soliton transmission under control of inline filter are studied with variational method. Then the analysis expressions of timing jitter and equations of DMS's transmission evolution are gotten. The analysis results indicate that the filter can increase the DMS's transmission rate and stabilize DMS transmission. At same time, there exist a optimal configure scheme of filter strength and parameters of DMS. If choosing the parameters (^k_f",C,B)reasonably, the pulse can be transmitted stable.

The design of a combined functional phase-only diffractive optical element, which can be utilized in intersatellite optical communication, is demonstrated. The adopted design method of this paper is based on the nonlinear phase transformation by the law of colour separation, Rayleigth-Sommerfield diffraction theory, direct search optimization method, and Genetic algorithm. This combined functional phase-only diffractive optical element (CPDOE) is motivated by the intersatellite optical communication (IOC). To simplify the structure of optical systems in IOC, the PDOE combined two different functions into the same one element. The two functions are the colour separation and spatial focusing of beacon laser and signal laser. With the utilization of this element the structure of IOC can be highly simplified and integrated. The motivation of being used in IOC also raises the demands of large numerical aperture, polarization insensitivity and high diffraction efficiency. In order to predict the large angle diffraction precisely, the diffraction between the back side of DOE and focusing plane is computed by the means of Rayleigth-Sommerfield (RS) diffraction theory, a more rigorous diffraction theory than Fresnel diffraction theory. To get a high performance of optimization, the direct search and Genetic algorithm has been adopted, and the initial phase value has also be synthesized by the nonlinear phase transform by law of colour separation. The presented design result showed a high performance of the CPDOE. The high performance CPDOE will play an important rule in IOC to increase communication rate as well as decrease volume and weight.

The effects of the PMD and filter control in DMS system are studied theoretically with variational method. Firstly, the DMS transmission of filtering control in constant birefringence fibers is analyzed by Varitional method, from which the evolving rules of characteristically DMS parameters are obtained. Secondly, the stability of DMS transmission and its timing jitter are investigated in the random varying birefringence fibers. At the last, the mean square of DMS timing jitter is gotten, which can be simplfied to no filtering DMS system and to linear transmission systems in considering the effects of PMD. The results reveal that the condition of stability of filter control in DMS system must be satisfied with the relationship of filter strength k_f < 2d_yP³ E_sol ÷ √πB⁴C² + 2_yP³ E_sol C and DMS's timing jitter can be decreased considerably with help of filter strength.

In this paper, the research interest focuses on the influence of buffer configuration methods on the performance of a shared WDM optical buffer. In order to resolve FDL contention, which often occurs in am shared WDM optical buffer, two improved buffer configuration methods, i.e., duplicated degenerate mode (DDM) and increased degenerate mode (IDM) are proposed. The performances of the proposed buffer configuration methods are evaluated by simulation experiments.

Due to lack of optical random access memory (ORAM), fiber delay lines (FDLs) is a good approach to contention resolution in optical packet switching (OPS). For studying of the performance of FDLs in content resolution strictly, a model of optical packet switching with feedback FDLs is established. The packet loss probability and the average delay time are deduced in theory. The results indicate that FDLs can improve the packet loss probability dramatically in low load but tardily in high load. Under the condition of low load, the difference of the average delay time with the different number of FDLs or input / output fibers is very small. In high load, the average delay time decreases with the increase of the number of input / output fibers and increases with the increase of the number of FDLs.

This paper addresses optimal multicast wavelength assignment problem in WDM network with limitedrange wavelength conversions (LWC). The optimal object of the problem is to minimize total number of LWCs required. We introduce the distortion constraint, which requires that the number of needed converters along the path from multicast source node to any member cannot be larger than the specified value. We solve the problem exactly by dynamic programming under session blocking policy and destination blocking policy, and present the experimental results.

Commercial and military interest in free-space optical communication (FSO) is growing due to the high bandwidths, portability and high security. FSO systems with proper radio-frequency backup can provide a viable solution for the last mile problem The communication reliability of FSO has been being pay attention to. In this paper, an innovation solution of light power equalization is put forward and some technologies such that enlarging the dynamic range of light power gain and realizing communication for FSO in the condition of all-weather are studied, a method to implement the power equalization is also given in term of system-embedded, some experiments are implemented with a system self-fabricated, too, the experimental results shown that a wide range power equalization system on FSO can effectively reduce the influences from the random variation of atmospheric medium and other factors on the communication link path.

Using dispersion compensation CBG, 2500km-10Gbps RZ and CSRZ transmission system on G.652 fiber is successfully demonstrated without electric regenerator and without Raman amplifier. At 2080km and 2560km, power penalty for RZ is about 3dBm and 5dBm (condition: RZ signal, BER=10^-12, PRBS=10^23-1); And power penalty for CSRZ is about 1dBm and 3dBm. The result indicates it is superior to the system using CSRZ signal under same condition.

In 1968, physicist V.G. Beselago pointed out that there always exists negative refractive index in materials when their electric constant and magnetic permeability μ are both negative, and therefore the propagation properties of the EM wave in such materials behave distinctively from those in media with positive refractive index (RI). In this paper, we construct two types of one-dimensional photonic crystals made up with positive and negative RI materials and investigate their nonlinear optical properties. One type of photonic crystals consists of materials with both negative linear and positive third-order nonlinear RIs, the other are mainly made up with a Bragg micro-cavity constructed with both linear positive and negative RIs which contains Kerr media. We show that both types of photonic crystals exhibit bistable switching characteristics. Moreover, as compared with the crystals involved pure positive RI, our photonic crystals possess even lower threshold valve in optical switching.

We consider hierarchical optical networks with different multigranular OXCs (concatenated and feedback). We compare and summarize various hierarchical routing models. We propose an intelligent two-step tunnel-reusing method (TSTR) to design multigranular meshed optical networks.

Performance of all-fiber fast frequency hop optical code division multiple access system (FFH-OCDMA) is studied in case parametrical difference exists between the fiber Bragg grating (FBG) arrays based encoder/decoder, i.e., the relevant gratings' Bragg wavelengths or relative positions are not strictly consistent. If there are such parametrical differences, some optical pulses with specific frequencies will be lost partially, or even entirely, which reduces the peak value of the self-correlation decoding output. The multi-users simulation model of FFH-OCDMA system is presented amply in the paper, which consists of these components: signal source, encoder, decoder, fiber channel, receive detector, and data analysis module. Simulation result is presented and discussed. It shows that the system performance drops down sharply with a larger wavelength inconsistency or with a larger position relative offset, especially when the number of FBG is smaller. It is also concluded that performance degradation can be alleviated by employing longer FBG array, but with the expense of higher system complexity.

In the paper, a high speed magneto-optic switch based on the Faraday Effect is designed and analyzed. The optic switch adopts Faraday rotator, nanosecond impulser, and high speed magnetic field, so it is characterized by no moving parts, low transmission loss and polarization insensitive, low optical insertion loss etc. Using the polarization and Faraday Effect of magneto-optic crystal, the magneto-optic switch can hold the function of all-optical switching, which is needed in all-optical communication networks. As the first part of this paper, a design scheme of optical route in high speed magneto-optic switch and its experiment analysis will be discussed. Good avalanche effect of transistor 2N5551 is adopted to generate nanosecond pulse signal and then to drive the high speed magnetic field. Shown by the experiment data, the rising time of the impulse about 10ns, the amplitude of the impulse about 10~60V are available on the Output end from the nanosecond impulser, which can be used as driving current pulse of Faraday rotator. By using the relationship between the polarization plane rotate direction of polarization light and magnetic direction, the Faraday rotator is designed. It's unique double magnetic field and externally applied static magnetic design can greatly speed the excitation time of the internal inductive magnetic field and shorten the switching time of magneto-optic switch.

Brillouin distributed optical fiber sensing system based on spontaneous Brillouin intensity measurement may measure temperature in the optical fiber, due to the dependence of spontaneous Brillouin scattering intensity on temperature. In this paper, a 4.25km distributed optical fiber temperature sensing experiment system was demonstrated. Using the experiment system, we obtained the traces of spontaneous Brillouin scattering intensity as the sensing fiber was not heated and heated respectively. By normalizing the trace of the fiber being heated to the trace of the fiber not being heated, the intensity change of spontaneous Brillouin scattering corresponding to temperature along the fiber has been obtained. And there was a good agree between the result and the heated section of the fiber.

Wobbling phenomena of spatial soliton in an inhomogeneous medium is investigated. By means of effective-particle approach method, we analyze the dynamics of the solitons. The results show that solitons wobble in transverse direction during propagation. The normalized width of the waveguide play a key role in determining the wobbling periods of solitons. Furthermore, the propagation dynamics of the solitons are simulated numerically and a good agreement is obtained between the analysis and the numerical results. This phenomenon may be used in all-optical router, switcher etc.

It is investigated the optical transmission properties of the Bragg microcavity sandwiched between Bragg reflectors which are composed of alternately arranged different dielectrics with positive and negative refractive indices. The defect layer thickness, incident angle of incident wave, quasi-periodicity of the Bragg reflectors and dispersion of the negative-refractive-index dielectric all have influence on the transmission spectra. Comparison of optical bistable characteristics is presented for the following three nonlinear Bragg microcavities. The first has Bragg reflectors with periodic structure containing non-dispersive negative-refractive-index dielectric; the second has the same structure as the first, however, the negative-refractive index dielectric is dispersive; the third also has periodic structure, however, where the Bragg reflectors are composed of alternately arranged dielectrics with different positive- refractive-indices. It is analyzed influence of the incident angle on the bistability of the nonlinear Bragg microcavity containing non-dispersive dielectrics.

Time-frequency analysis theory was used in this work to analyze chirped optical pulse signals produced in directly modulated semiconductor lasers or optical propagation channel. The partial characters of the chirped signals can be depicted at time domain and frequency domain simultaneously using these methods. For chirped gaussian pulse signals which are commonly used in optical communication, Short Time Fourier Transform and Wavelet Transform as the examples of linear time frequency transform and Wigner-Ville distribution as for nonlinear time-frequency transform are analytically presented. In the end, both advantages and disadvantages of the time-frequency analysis methods are compared.

The three allocated schemes of dispersion managed soliton transmission systems are proposed on standard single-mode fiber. Adopting variational method, the timing jitters induced from the perturbations of amplifier spontaneous-emission (ASE) noise and across-phase modulation (XPM) in multiplexed systems are studied. The results have shown that the dispersion management strength S must be satisfied 2