The high tech bubble was inflated by myths of astronomical Internet traffic growth rates. Yet although these myths were false, Internet traffic was increasing very rapidly, close to doubling each year since 1997. Moreover, it continues growing close to this rate. This rapid growth reflects a poorly understood combination of many feedback loops operating on different time scales. Evidence about past and current growth rates and their sources is presented, together with speculations about the future. The expected rapid but not astronomical growth of Internet traffic is likely to have important implications for networking technologies that are deployed and for industry structure. Backbone transport is likely to remain a commodity and be provided as a single high quality service. It is probable that backbone revenues will stay low, as the complexity, cost, and revenue and profit opportunities continue to migrate towards the edges of the network.

Global communications traffic continues to grow, with packet switched traffic beginning to dominate the network. In the long run, new network optimizations and fiber optic system technologies are going to be needed to meet this need.

Multi-wavelength transmitters and receivers are the key elements for point-to-point WDM links. The functionality of the links is increased by the addition of optical add-drop multiplexers (OADMs) and WDM networks are formed by using optical cross-connects (OXCs). For a successful monolithic integration of these WDM components, the number of elementary components from which they are made has to be small, in order to keep the technological challenges manageable. In this paper we will identify the building blocks and show how they are used.

Recent laboratory experiments have demonstrated that distributed Raman amplification, advanced modulation formats, optimized dispersion maps, and forward error correction are key technologies for 10-Gb/s and 40-Gb/s DWDM terrestrial transmission over 2000 to 6000 km. The transmission fiber's Raman gain efficiency and dispersion properties are thus important parameters. Future high-bit-rate, high-capacity installed systems will require advanced transmission fibers to extend their reach to at least 2000 km, a distance also specified by a high-profile U.S. government optical networking project. This paper will address a number of the enabling fiber properties, including dispersion, dispersion slope, Raman gain efficiency, and polarization mode dispersion. In addition, several recent experiments will be reviewed, including demonstrations of high-spectral-efficiency terrestrial transmission at 10 Gb/s and 40 Gb/s over 4000 km and 3200 km, respectively, and 10-Gb/s transmission over 2400 km using 200-km spans.

In this paper we investigate the transmission performance of 16Nx10 Gb/s, 4Nx40 Gb/s and Nx160 Gb/s DWDM systems with equal spectral efficiency (0.8 bit/s/Hz). For a comparison between different systems optimized filter, fiber settings as well as optimized dispersion compensation scheme have been used. It is shown that if the considered amount of photons per bit is constant all investigated systems show similar transmission characteristics.

A general modeling and simulation strategy suitable for the fast and accurate analysis of a fiber-optical WDM system is presented, that may also include multi-span systems. Noise and fiber dispersion are considered as well as nonlinear effects like four wave mixing, self-phase modulation and cross-phase modulation. Furthermore, amplified spontaneous emission noise of the optical amplifiers and polarization-dependencies (e.g. PMD) are taken into account. Performance evaluation by means of eye patterns, spectral power densities, optical signal-to-noise ratio, the Q-Factor and the bit error rate are addressed. An analysis of the degradation effects against the position within the fiber is shown to get a better overview of the fibers behavior. An improved Split-Step algorithm is outlined as a fast alternative to supplement the FFT calculation within the fiber. A parameter variation, which can also be influenced during the simulation by the user, is presented in order to get an overview of the parameter space. Different modulation formats are taken into account, e.g. return-to-zero, non-return-to-zero and differential phase-shift keying. Both the separated channels and the total field approach are demonstrated. In the separated channels approach the different nonlinear effects can be switched on and off independently for detailed studies of inter-channel effects. Based on this work, a complete design environment (PHOTOSS, The Photonic System Simulator) has been developed. This simulation tool has been tested extensively by several industry partners. Simulation examples are presented here e.g. for PMD simulations.

In this paper cross-phase modulation (XPM) and its impact on multispan NRZ-modulated wavelength division multiplexing (WDM) systems is examined. An analytical model is presented to assess the signal degradation. The impairments due to XPM are related to a Q-factor. The presented formulas are examined for different dispersion compensation schemes and also mixed-fiber systems. The analytical models are verified by system simulations employing the split-step Fourier method. To determine the impairments due to XPM the pump-probe model is used, which assesses the distortion of an unmodulated probe channel. It is shown that this model is applicable to NRZ systems. Existing analytical models are extended for certain conditions, e.g. given dispersion compensation schemes such as the distributed under-compensation scheme (DUCS) or the full-inline and optimized post compensation scheme (FOCS). Furthermore, a formula will be suggested to estimate the degradation of performance, if further WDM channels are added to an existing system.

Chromatic dispersion and polarization mode dispersion (PMD) are the most relevant disturbances in optical high speed transmission systems. The influence of chromatic dispersion can be overcome by different methods such as dispersion compensating fibers or fiber gratings. Because of its stochastic behavior the struggle against PMD is much more troublesome and essential for bit rates above 10 Gbit/s. The paper describes the influence of polarization mode dispersion derived from basic principles of transmission. It gives some physical understanding and simple derivation of the characteristic equations of PMD. One distinguishes between first and second order PMD. An optical fiber link showing first order PMD only can be modelled by a waveguide with time varying birefringence. If the birefringence additionally is frequency dependent the fiber will show higher order polarization mode dispersion. PMD compensators can be divided in a similar manner: compensators considering first order PMD only and compensators including PMD of higher order. Compensators operating in the optical domain or in the electrical domain have been proposed.

Operators of today's high-speed optical communications networks are required to deliver error-free transmission and continuously increasing bit rates. The complexity of synchronous optical networks (SONET, SDH) and high-speed data communications systems (Gigabit Ethernet) allows little tolerance for signal impairment. As a result, strict requirements on performance metrics such as synchronization, jitter, bit error rate (BER), and availability are all necessary. This paper reviews jitter and how it may affect the performance of optical communications systems. This will include discussion of important jitter parameters, sources of jitter, and test methods used to measure jitter in optical networks. Measurements of various jitter scenarios on OC-192 transmission will be presented, and the impact of optical signal impairments on OC-192 performance will be shown. Finally, the relationships between optical signal impairments, observed jitter, and 10GbE stressed-eye tests will be discussed.

We have developed a wavelength domain simulation approach that can estimate the dispersion penalty due to initial chirp. This method only requires estimation of two parameters, i.e. effective chirp duration and effective chirp peak value, from the measured time resolved chirp (TRC) data. It is very fast comparing to full time domain simulation since it does not require the calculation of the time domain signal propagation in fiber. The method is based on a simple phenomenological model that is capable to predict the dispersion/chirp-induced eye-closure penalty. Model can estimate other system performance indicators such as BER and Q factor by taking into account the effects such as ASE, thermal and shot noise, extinction ratio etc. We have compared our results with published experimental data and full time domain simulations and have observed very good agreement.

This paper surveys recent progress in fiber spinning technology for reducing polarization mode dispersion with the focus on theoretical understanding of fiber spinning mechanism and features of spun fibers, as well as their applications in fiber spin profiles designs. First, a brief introduction to the requirements of fiber PMD for high-speed optical fiber communications systems is given. Further discussion covers several spin technologies, including preform spinning and fiber spinning. Different theoretical approaches for modeling fiber spinning are reviewed with the emphasis on the coupled-mode formalism. The coupled-mode theory is then applied to different fiber spin profile designs, including constant, sinusoidal and frequency and amplitude modulated spin profiles. Main features of these spin profiles, such as PMD reduction dependence on fiber beatlength and sensitivity to spin parameters, are compared. Scaling properties of spun fibers, which show different behaviors from unspun fibers, are also discussed.

During the 1990s, a lot of promises were made concerning optical switching and the role it will play in future transport networks. A new optical layer based on Wavelength Division Multiplexing (WDM) and optical switching was envisioned. As the industry is going through a downturn, optical switching has not been deployed and the optical layer has not been realized in practice. In this paper, we look at the past and present of optical switching and try to explore its potential in future networks.

Optical burst switching (OBS) has been proposed in the late 1990s as a novel photonic network architecture directed towards efficient transport of IP traffic. OBS aims at cost-efficient and dynamic provisioning of sub-wavelength granularity by optimally combining electronics and optics. Optical bursts cut through intermediate nodes, i.e., data stays in the optical domain at all times, while the control information is signaled out-of-band and processed electronically. In contrast to optical packet switching, OBS aggregates and assembles packets electronically into bursts of variable length according to destination and QoS class at the edge of the network. This paper surveys current trends in OBS and discusses proposed solutions for burst reservation and scheduling, burst assembly, contention resolution and QoS provisioning as well as design and scalability of OBS nodes. Also, it looks at the question of the optimal burst size, which has been around from the very beginning of OBS, based on recent trends and results. On the one hand, burst size and burst characteristics are influenced by client layer traffic and burst assembly scheme. On the other hand burst size and burst characteristics have an impact on network performance and node architectures. Finally, consequences of burst durations in the microsecond and millisecond range are presented and compared.

Optical CDMA provides one avenue to sharing the vast bandwidth of optical fiber among a number of active users in a network. In this work we address two questions central to considering the areas of applicability of CDMA in optics: (1) How do we compare fairly optical CDMA codes of different families, and what do we conclude? (2) How do we compare optical CDMA performance with that of conventional optical WDMA? To answer the first question, we present a method of comparing optical CDMA codes of different families, sizes and weights. We outline why the traditional performance metric of bit-error rate versus number of simultaneous users is lacking and propose a different performance measure -- the peak throughput normalized with respect to the size of the code. This metric is used to show that optical-orthogonal codes (OOCs) with a weight of 4 perform best at low offered loads while OOCs with weight 5 should be used at higher offered loads. By applying the technique across different families of codes, we demonstrate that multi-wavelength OOCs (MWOOCs) perform better than both OOCs (by a factor of approximately 1.25) and asymmetric prime-hop codes (by a factor of approximately 3.5), over a wide range of offered loads. In answer to the second question, we present a comparison of the performance of optical code-division multiple-access (CDMA) systems with that of a wavelength-division multiple-access (WDMA) system. The multiple-access techniques are applied in a time-slotted broadcast local-area network. The utilization, defined as the throughput per unit of time-domain bandwidth expansion, and packet delay are used as metrics of performance. When more than seven wavelengths are available, optical CDMA systems using asymmetric prime-hop codes and all-optical signal processing are shown to have higher peak utilization and lower corresponding delay than a WDMA system with the same number of wavelengths. When the encoders/decoders operate at the chip rate, the utilization of optical CDMA exceeds that of WDMA at high offered loads; however, the peak utilization of the WDMA system is still superior.

Versatile OC-based MPLS for circuit-, burst-, and packet switchings is presented. The OC-label is an identifier attached to the data, of which information is mapped onto an optical code, a sequence of optical pulses. The OC-label recognition is performed by the optical correlation between an incoming OC-label and the OC-label entries. Unique to the OC-label processing is that it is analogue operation entirely in optical domain, and no optical logic operation is involved, therefore, the processing speed is only limited by the propagation delay in the optical correlator, which is a passive optical device. This is a key to the ultrahigh-speed processing capability. The OC-label switching technology will create a new paradigm for the next generation photonic networkings.

There is an increasing interest in performing many key networking functions in the optical domain to achieve bit rate transparency. Optical header processing is one such key function that may enable fast reading and forwarding of optical packets in the future all-optical packet-switched core network. Many of these optical header processing functions are enabled through the use of all-optical logic gates. The logic XOR gate is of key importance in decision and comparator circuits. A novel architecture of an N-bit logic XOR gate based on a Mach-Zehnder interferometer with feedback is proposed and its performance evaluated by means of simulations. Basically, this architecture consists of an integrated semiconductor-optical-amplifier-based Mach-Zehnder interferometer (SOA-MZI), an optical pulsed control signal, a differential transmission scheme for the input data sequences, and a feedback network. The simulation results show error-free operation at 40 Gbit/s for 16-bit-length words with extinction ratio values better than 16 dB. Furthermore, simulation results of the data power threshold needed for obtaining error-free operation as a function of the peak power of the control pulses are also presented, showing an optimum operating point at about 8 mW. An important application for the proposed SOA-MZI architecture is label processing directly at the optical domain in high-speed all-optical label swapping networks.

A new system for generation, switching and routing of optical packets in photonic networks, having 2x2 optical transparent nodes, is described. The optical packets are constituted of a frequency tone header in the low RF range (MHz), and a digital payload which can be of different rates or formats, in the range 1 to 5 Gb/s. Packets can have time frame between 2-6 μs, with separate fields for header and payload; the header field being much shorter than the payload field. The optical packet switch node includes blocking, routing and drop functions, controlled by electronic logic circuits, processing only the header information. Switching is performed on a packet-by-packet basis; and packets are routed according to header frequency allocation, which relates to node output ports and packet destination node addresses. Processing times for header recognition and packet switching are 2 μs. The digital payload remains untouched throughout the network, delivering high-bandwidth to final users, with very low latency in the network, and very small packet loss. This system is designed for application in metro-access next-generation optical networks (NGON), one example being interconnection of high-capacity wireless base stations.

The paper addresses the migration path and trade-offs involved in the evolution towards intelligent optical networks that allow scalable and flexible transport of a range of client formats and services. Disruptive all-optical switching and ultra-long reach WDM transmission technologies enable all-optical networks without opto-electronic conversion along paths that can span thousands of kilometers. Optical transport networks coupled with advances in distributed routing and signaling mechanisms are enabling the deployment of Automatically Switched Optical Networks (ASON). The paper discusses in detail the network architecture, optical switching capability, and network management and control architecture involved in the migration from existing and widely deployed WDM optical point-to-point systems to ASONs. The paper reports the findings of the Global Seamless Network (GSN) demonstrator at Deutsche Telekom, in which the concepts of intelligent optical transport networks are being demonstrated and evaluated. Further, a special emphasis is devoted to the question of managing survivability in ASONs. Simulation results comparing different approaches are presented, which illustrate the trade-off between dimensioning and performance of resilience mechanisms. The multi-layer integration between optical and higher layers is addressed from both a control plane and survivability perspective. Scenarios with different clients of the optical layer such as SDH and IP/MPLS are analyzed.

Future multi-terabit/s optical core networks require optical technologies capable of managing ultra-high bit rate OTDM/DWDM (optical time division multiplexing/dense wavelength division multiplexing) channels at 160 Gbit/s or higher bit rates. The key functionalities in ultra-high speed network nodes are all-optical wavelength conversion, 3R-regeneration and demultiplexing of OTDM signals. Advanced optical networking techniques (optical add-drop multiplexing and optical routing) are studied in simulations and their performance evaluated considering 160 Gbit/s OTDM/DWDM channels. Performance comparison results for both OADM (optical add-drop multiplexer) and OXC (optical cross-connect) node networking functionalities are shown considering different technologies: semiconductor-optical-amplifier-based symmetric Mach-Zehnder interferometers (SOA-MZI) for wavelength conversion, signal regeneration and demultiplexing, electroabsorption-modulator-based demultiplexers, and wavelength converters based on four-wave mixing in dispersion-shifted fiber. The simulation results show that the SOA-MZI is a promising technology for all-optical signal processing in network nodes mainly due to its signal regeneration capability. At ultra-high bit rates, however, the relaxation time of SOAs considerably limits the operation. A solution to mitigate this problem is to use a differential scheme at the input of the device. Error-free wavelength conversion, signal regeneration and demultiplexing of 160 Gbit/s OTDM signals employing a SOA-MZI with a differential scheme is demonstrated by means of simulations. Furthermore, the parameters of this architecture are optimized to obtain the best performance for each optical networking functionality in OADM and OXC network nodes.

Equipping all nodes of a large optical network with full conversion capability is prohibitively costly. To improve performance at reduced cost, sparse converter placement algorithms are used to select a subset of nodes for full-conversion deployment. Further cost reduction can be obtained by deploying only limited conversion capability in the selected nodes. In this paper, we present limited wavelength converters placement algorithm based on the k-Minimum Dominating Set (k-MDS) concept. We propose three different cost effective optical switch designs using the technologically feasible non-tunable optical multiplexers. These three switch designs are Flexible Node-Sharing, Strict Node-Sharing and Static Mapping. Compared to the full search heuristic of O(N³) complexity based on ranking nodes by blocking percentages, our algorithm on one hand has a better time complexity O(R.N²) - R is the number of disjoint sets provided by k-MDS; and on the other hand avoids the local minimum problem. The performance benefit of our algorithm is demonstrated by network simulation with the U.S Long Haul topology having 28 nodes (R is 5). From the optical network management point of view, our results also show that the limited conversion capability can achieve performance very close to that of the full conversion capability; while not only decreasing the optical switch cost but also enhancing its fault tolerance.

It is anticipated that future core networks may evolve to Optical Cross-Connect (OXC) based architectures where switching is carried out at the granularity of wavelength-based circuits. While the switch fabric is an important component of an OXC, the number of ports is typically a more significant contributor to its cost. Reducing the number of ports (or number of wavelengths per link) will also result in lower transmission-systems costs. We investigate the effect of incorporating some packet switching functionality in the OXC and try to quantify the bandwidth savings that can be gained by doing so, under certain assumptions about the network and the traffic characteristcs. Results show that taffic grooming and statistical multiplexing gain, which are introduced by the packet handling functionality, lead to better channel utilization. This can result in lowering the number of channels and ports and in reducing the OXC size and cost.

Although the Optical Transport Network, based on technologies such as Wavelength Division Multiplexing and Optical Cross-Connects, offers tremendous transportation capacity, its management requires frequent manual intervention. However, as the traffic pattern offered to today's transport networks is subject to continuous changes due to the Internet traffic dominance, an optical transport network with a smart, automatic and real-time control system, denoted as Intelligent Optical Network (ION) or Automatic Switched Optical Network (ASON), is desired by network operators. Duly and correctly retrieving the changing traffic load information is a very important factor for the successful deployment of an ION. In this paper, we discuss the influence of the observation window size used for collecting the traffic load information, on the performance of an ION. By comparing the performance of an ION using different traffic observation window sizes, we show that a smaller observation window harms the network stability; while a too large observation window worsens the network reliability. We demonstrate that a suitable traffic observation window size improves the offered Quality of Service (QoS) by reconfiguring the logical layer network at the right time and in the right way.

As more and more traffic is transported over communication networks, network survivability becomes a key issue in network design and planning. In this paper first the need for deploying network recovery techniques at multiple layers is motivated. Then the efficient coordination of these network recovery techniques is studied. Not only static but also dynamic multi-layer survivability strategies are presented and studied in this paper: dynamic multi-layer survivability strategies benefit from the ability of the underlying transport network to provide switched connection services in order to allow reconfiguring the logical network at the time of a failure. As not only the traffic volume keeps growing, but also more and more services with distinct reliability requirements are deployed, the benefit of differentiating the multi-layer survivability strategy per traffic class is investigated in this paper. Whereas the case studies in this paper focus on the cost-efficiency of the presented strategies and techniques, also a broader theoretic discussion is given on these techniques.

Network security is becoming a very sensitive and important topic for equipment manufacturers and network operators. In transparent optical networks, security is even more complex since the optical signals are not regenerated as in opaque networks and, therefore, the faults and attacks at the physical layer are more difficult to detect and isolate without significantly affecting the overall network performance. In this paper we define Failure Management as the prevention, detection, and reaction against failures. Failures are defined as the interruptions of the normal functioning of the network and comprise faults (accidental interruptions) as well as attacks (intentional interruptions which can be performed by service disruption or eavesdropping). Our work deals with a solution to detection of failures in transparent networks. For this purpose we have extended a Fault Location algorithm developed for opaque optical networks to be used in transparent networks and be able to also locate attacks. The proposed algorithm is called Transparent Failure Location Algorithm (TFLA). The first part of the extension is based on the study of other optical network elements such as Optical Add/Drop Multiplexers (OADMs), Optical Cross-Connects (OXCs), wavelength converters, Optical Line Terminators, etc. The vulnerability of these elements depends on their architecture and/or fabrication technology and, therefore, different attacks can be considered. A classification of these components based on the masking and alarming properties is proposed. The second part of the extension is based on the monitoring equipment that may be available in transparent networks. The TFLA was applied for the case of a transparent ring of the Pan-European network.

All-optical networks (AON), using wavelength division multiplexing (WDM), have become attractive candidates for building wide area networks (WANs). Finding a logical topology and a routing scheme making optimum use of network resources is a challenging task. A complication is that the survivability of AON has become an important issue. In designing a fault tolerant WDM network, the primary lightpaths, the corresponding backup lightpaths, and the routing scheme have to be determined simultaneously in such a way that network resources are used in an optimum manner. In this paper we first develop an Integer Linear Formulation (ILP) for designing a fault-tolerant logical using shared path protection. The objective is to design a survivable logical topology and a routing over that topology in such a way that the overall congestion is minimized. This formulation can be solved to give us the optimum logical topology for small WDM networks. However, for larger networks, this approach becomes infeasible due the large number of constraints and integer variables. For such networks, we outline a simple heuristic algorithm to find a feasible logical topology.

Networks add recovery to deliver reliable service between a source and destination. Reliable service is an uninterrupted data transfer between nodes, even if certain links or nodes are faulty. Recovery processes in multilayer networks might operate at the same time when faults propagate between layers during a common-cause failure. We propose an architectural model, that represents point-to-point networks as a grid, where the decisions of recovery can be represented as the replacement of ports. Since a connection is a path through the grid, we can easily determine if multiple connections are repaired. Repairing connections is the goal of multilayer recovery, even if they are repaired at different layers. Since the operations of recovery are the result of preplanned provisioning, a model-checking algorithm could be implemented to check for conflicts in the provisioning. As a first step, this paper decouples the protocols for implementing the actions of recovery from an actual resource-based model. Therefore, no signaling between nodes is treated. Since the provisioning of recovery processes will create decisions of recovery after a catastrophic common-cause failure, we can map the configurations that satisfy goals into provisioning commands sent to recovery processes.

Relaxing the Optical Signal-to-Noise Ratio (OSNR) requirements of modern Wavelength Division Multiplexing (WDM) transmission systems/networks is essential due to the introduction of high-loss Optical Network Elements (ONE), the large channel numbers, and the very high channel-speeds. Forward Error Correction (FEC) codes are currently used in many applications, but the problem of the bit-rate increase caused by the redundancy will worsen as the channel-speed increases. This paper presents a novel error-controlling scheme that does not increase the bit-rate while offering OSNR relaxation. Error control is achieved by the two kinds of receiver proposed here, both of which are characterized by the diverse detection of dual-mode modulated signals, “consensus” logic for error detection, and multiple threshold selection for data recovery. After describing the “basic” receiver, an enhanced version is presented. Analyses based on the classical Gaussian noise approximation show the performances of these receivers. Experimental results show fair agreement with the analysis results, despite the simplifications used in the analyses. Experiments on the basic receiver show 3dB gain, and 9dB gain was observed with the addition of the standard RS(255,239) FEC code. 4dB gain is observed experimentally from the enhanced version of the receiver, and 10dB gain with the addition of RS(255,239).

We report advanced modulation formats in transmission systems operating for 40Gb/s DWDM for multi-Tb/s photonic transport networks including (1) efficient RZ, RZ-CS, RZ-CS VSB for optimizing dispersion tolerance in linear and nonlinear operating regimes (2) disperison management and compensation techniques (3) equalization of gain by lumped and distributed amplification. Simulated and some experimental results are given.

Carrier-suppressed return-to-zero modulation (CS-RZ), in which there is a 180-degree phase shift between successive pulses, reduces the effects of intrachannel impairments. However, there are a number of variants of CS-RZ that differ in their method of generation, their bandwidth requirements, and their performance. It is shown that a recently proposed technique, in which the RZ signal is generated by filtering a CW signal that is square wave phase modulated (termed CWSW), performs comparably or better than alternative techniques. Relative to a modified duobinary system that results in alternate mark inversion( AMI), CWSW achieves better performance in systems with small dispersion, but slightly poorer performance in systems with larger dispersion and dispersion compensation. The small improvement in performance of AMI relative to CWSW in this latter case is achieved at the expense of requiring a larger transmission bandwidth and more complex transmitter. The physical basis of the impairments in these systems, a peak intensity enhancement phenomenon in CWSW that counters the effects of dispersion, and factors affecting the bandwidth of alternative techniques is discussed.

A new version of Duobinary-Polarization Shift Keying scheme is proposed. The Duobinary electrical filter has been moved to the receiver side. Optimization of both transmitter and receiver electrical filters gives a receiver sensitivity improvement while maintains both chromatic disperison and non-linear effects tolerance. The main reason of this sensitivity increase is the reduction of the timing jitter given by our proposal.

Novel modulation formats have been a key issue in many recent reports on 40-Gbit/s applications. In particular, coherent transmission technologies including the differential phase shift keying (DPSK) format in combination with balanced detection have merited a lot of attention, because of its robustness against fiber nonlinearity induced impairments in DWDM applications. We have proposed several novel transmission formats based on amplitude shift keying technology including carrier suppressed return to zero (CS-RZ) and duo-binary carrier suppressed return to zero (DCS-RZ) formats. We have also proposed CSRZ-DPSK formats based on phase shift keying (PSK). In this paper, we discuss the remarkable relation between these advanced signal formats that can be afforded by using the spectral mode splitting (SMS) scheme. All these modulation formats can be interrelated via the SMS scheme, which transforms one signal format into another via spectral filtering. Furthermore, we discuss the transmission performance of these signal formats and show how to improve their performance by employing the SMS scheme. We can achieve both higher tolerance for fiber dispersion and higher allowable launched optical power by employing the SMS scheme. The SMS scheme also provides us with useful applications such as chromatic dispersion monitoring.

The impact of phase noise, chromatic dispersion and non-linear effects on a 10 Gb/s Optical Differential Quaternary Phase Shift Keying (oDQPSK) system has been evaluated by means of computer simulation. Both single channel and multiple WDM channels have been considered. The results show that these systems present a high spectral efficiency that allows a reduced channel spacing as low as 25 MHz. Furthermore, this effect leads to an increase on dispersion tolerance together with a non-linear effects robustness due to the signal constant envelope characteristics. Finally, some relevant results have been found concerning the laser linewidth requirements (less than 10 MHz). Simulations show that phase noise can be the limiting factor of oDQPSK systems.

Fiber optics is now widely deployed in data networks inside of buildings and in the long haul telecom networks. But tying these networks are copper based transmission lines in the metropolitan areas. As fiber is able to take over in the metropolitan access networks a uniformly high bandwidth, low latency network emerges. Technically speaking this is the most challenging area for optical communications technology. It requires low cost, sometimes extended temperature, small size and power and very often WDM on top of all these other constraints because laying new fibers for this network is prohibitively expensive. Use of WDM in these markets allows use of lean fiber plant simultaneously with service separation by wavelength. Thus different protocols and end user needs are flexibly met on a pay as you deploy and use basis. This paper draws a number of simple metro access rings each with different bandwidths and protocols and technologies. Costs as well as performance will be cited on a per node basis.

CATV (Community Antenna Television) networks were developed several decades ago to deliver a few channels of analog television to a fraction of the homes in a small area. Over the intervening years, these networks have evolved into true communications networks supporting digital video and the delivery of different targeted services to homes and business covering a much larger area. Supporting targeted services and spanning the longer distances has required the application of several technologies and the development of new network architectures. These new services require the use of technologies such as WDM. First, the unique characteristics of CATV networks are discussed. Examples of WDM-based network architectures that can support targeted data services via CATV cable modems and Video-on-Demand (VOD) are also presented.

The use of photonics for phased array antenna (PAA) control is a vital area of applied research. In particular, the potential impact of compact, low cost, photonically controlled PAAs for large scale global applications, such as wireless communications, is a major business growth area. This paper presents an overview of technology approaches for photonic delay lines (PDLs) as well as the challenges. Polarization based switched PDLs using two-dimensional (2D) liquid-crystal devices are discussed. In particular, 2-D ferroelectric liquid crystal devices are used as polarization switches in PDLs to control and process optical signals. The experimental demonstration of a multichannel 7-bit PDL system for a wideband PAA such as the Navy's advanced Aegis radar system is presented. Finally, we present novel approaches for hardware and control complexity reduction. One hardware compression approach is based on ternary PDL designs and wavelength division multiplexing. The other approach is based on the Dingel-resonator which functions as a novel prism-like optical device using Gires-Tournois Resonator (GTR) whose front surface mirror is coated with pre-defined spatial reflectivity profile.

The atmospheric optical links (FSO) in visible and infrared wavelengths constitute an interesting alternative to creation of new transmission channels for the cordless phone, data-processing networks and high definition television. One finds a choice of varied manufacturers and they propose products whose performances are characterized by a raised rate of transmission, from 2 Mbps to 10 Gbps. But the announced ranges are very important, from 100 to 10,000 meters, in spite of the fact that many manufacturers try to indicate the possible ranges according to time, these indications completely miss standardization and are hardly exploitable because, generally, it is very difficult to know the percentage of time during which a value is reached or exceeded. Availability and reliability of a FSO link depend on used systems but also on climatic and atmospheric parameters such as rain, snow or fog. Library search underlined the lack of reliable data to be able to lay down, in a precise way, the statistical availability of such links, like one usually does for the radio transmission. Before to implement an effective FSO links, we need to know their availability and their reliability. It is the purpose of our study. Its finality is a software which integrate (1) Results of a library search (geometrical attenuation, aerosols, scintillation, environment light, etc), (2) English and French integrated surface weather data, hour per hour, over several years (1995-1999). The result is the presentation of this software, "QoSLight" (Quality of Service Light), making it possible to predict; starting from the data of equipment (power, wavelength, receiver sensibility), geographical situation of a site in France or England (geographical coordinates, altitude, height/ground) and climatic and atmospheric parameter (relative humidity, ground rugosity, albedo, solar radiation, etc) the availability of a FSO link for the following period (year, the most unfavourable month, 8am to 8pm period and 8 am period. The interruption probabilities for each type of attenuation are also mentioned (aerosols, scintillation, ambient solar light, rain, snow, etc).

Given a set of lightpath connection requests in an all-optical wavelength division multiplexed (DWDM) network, lightpaths are designed and the wavelength channels are assigned so that the scheme minimizes on average the blocking of existing and provisional requests. This paper proposes a routing and wavelength assignment scheme for DWDM long-haul optical networks that includes routing, assignment and reservation of different wavelength channels operating under the G-MPLS environment. The modeling is implemented under the framework of an object-oriented modeling platform OMNeT++. Network performance tests are evaluated for a long-haul terrestrial fiber network composed of 12 gigaPoPs with an average degree of 3. Traffic blocking of lightpath requests are examined with the average link utilization in the network employing variable number of wavelength channels. Two principal algorithm for wavelength assignments, the first-fit and random types, are studied and proven that the random scheme offers significant better performance than that of its counterpart.

With the rapid growth of the Optical Internet, high capacity pipes is finally destined to support end-to-end IP on the WDM optical network. Newly launched 2D MEMS optical switching module in the market supports that expectations of upcoming a transparent optical cross-connect in the network have encouraged the field applicable research on establishing real all-optical transparent network. To open up a customer-driven bandwidth services, design of the optical transport network becomes more challenging task in terms of optimal network resource usage. This paper presents a practical approach to finding a route and wavelength assignment for wavelength routed all-optical network, which has λ-plane OXC switches and wavelength converters, and supports that optical paths are randomly set up and released by dynamic wavelength provisioning to create bandwidth between end users with timescales on the order of seconds or milliseconds. We suggest three constraints to make the RWA problem become more practical one on deployment for wavelength routed all-optical network in network view: limitation on maximum hop of a route within bearable optical network impairments, limitation on minimum hops to travel before converting a wavelength, and limitation on calculation time to find all routes for connections requested at once. We design the NRCD (Normalized Resource and Constraints for All-Optical Network RWA Design) algorithm for the Tera OXC: network resource for a route is calculated by the number of internal switching paths established in each OXC nodes on the route, and is normalized by ratio of number of paths established and number of paths equipped in a node. We show that it fits for the RWA algorithm of the wavelength routed all-optical network through real experiments on the distributed objects platform.

Optical networking, with its almost unlimited bandwidth, is the only technology that can support communication new applications with high demand for bandwidth such as remote visualization, teleimmersion, collaborative e-science etc. In the optical networks presently deployed, each link is optically isolated by transponders doing O/E/O conversions. But these transponders are quite expensive and lack transparency; hence there is a strong motivation to use large all-optical networks. In all-optical networks, all routes need not have adequate signal quality. To provide reliable and transparent service, there is a strong interest in the industry for routing protocols that take into consideration the impairments of optical layer that degrade quality of the signal. We present a new routing protocol that uses information about the quality of signal to select paths in an all-optical network. Our simulations confirm that this new protocol reaches higher network utilization compared to existing ones.

This paper presents a policy-based management framework for GMPLS-enabled optical networks. First, the ASON (Automatic Switched Optical Networks) and GMPLS (Generalized Multi-Protocol Label Switching) architecture are presented. We will explain how the policy plane relates to the control and management plane, and the role it plays in the dynamic configuration of GMPLS optical paths. We will present a framwork based on an admission control protocol COPS) Common Open Policy Service) defined by the IETF (Internet Engineering Task Force) and on an Information Model based on the CIM (Common Information Model) model defined by DMTF (Distributed Management Task Force). This framework allows interworking between the policy and control plane, and provides a comprehensive policy provisioning setup for optical networks. An information model extended from CIM for managing dynamically configured optical networks is presented.

To obtain the partial information of network's resources, which indicates what fraction of each link's bandwidth, is currently used by working paths, and is used by protection paths, is feasible in IP over WDM networks using Multi-Protocol Label Switching (MPLS) technology. Based upon this partial information scenario, this paper proposes a new integrated dynamic routing algorithm of path-based protection design for IP over WDM networks. For working paths selection, the algorithm makes a tradeoff between the routing costs and the corresponding routing path length. While for protection paths, a tradeoff between the sharing of the protection paths and the associated routing path length is made. We find that the developed routing scheme is beneficial to achieve a good link load balancing. Simulation results show that the developed algorithm performs well in terms of performance metrics such as the number of rejected requests and the network link load balancing.

In IP/MPLS/WDM networks, routing, resource allocation are usually conducted separately in IP/MPLS layer and WDM optical layer, which often suffers from low resource utilization and low survivable performance. In this paper, under the integrated dynamic routing scheme of IP/MPLS/WDM, we compared and analyzed three different routing strategies VFPL, PFVL and MRRF, which differs in whether choose to conduct routing upon the virtual topology first or upon the physical topology first when an LSP connection request arrives. Via simulation, the performance of these integrated routing schemes in terms of blocking probability are presented.

In this paper, we present a scheme for recovering 10 GHz clock from 40 Gb/s and 80 Gb/s time division multiplexed (TDM) return to zero (RZ) data stream. The proposed clock recovery is successfully demonstrated using an electrical phase locked loop (PLL). The jitter of the recovered clock is estimated to be around 50 fs. The key part in the proposed clock recovery circuit is a LiNbO₃ Mach-Zehnder modulator which is shown to be highly effective in optical to electrical down conversion.

In the state of polarization principle, a component is cut in a lot of trunks. Each trunk is characterized by vectors linked to local PDL and PMD values and by the Pauli's matrix. This vectorial approach is used to determine the complete characteristics of an optical assembly as a function of PDL and PMD vectors of each trunk. We particularly put in evidence the output PDL vector evolution versus the wave time constant, (e.g. versus the frequency). A study of the PDL maximum and minimum evolution (position and value) is also achieved. This study shows the posituons of the PDL maximum and minimum values are constant for the highest wave time constant values. This point out that the PDL maximum (minimum) value is obtained when the PDL and PMD vectors of each trunk are parallel (anti-parallel respectively). An evolution of this values and positions is proposed for the lowest wave time constant values. For a sufficiently high frequency, it is possible to demonstrate the assembly had an almost null PDL.

The aim of this work is to present a customized simulation tool for analysis and design of Sagnac loop TOAD's. Basically, we present a fast and versatile tool to emulate the behavior of some key design parameters like those found on non-linear elements, splitters, fiber delay line among others. After proper optimization, we use two different wavelengths for the control and signal hyperbolic secant pulses. Additionally, a band-pass filter is used to guarantee the transmission of signal pulses only. Furthermore, we analyze a bit to bit all optical demultiplexer, where one channel is extracted from a 64 bit frame at 10 Gbps transmission rate, and extend the implementation to work at 100 Gbps. SNR better than 22 dB are achieved with control pulse energy on the order of 200fJ and SOA bias current of 150 mA. A degradation of SNR of about 17 dB was verified for a variation of splitter coupling factor around 20%.

In this paper we implement and characterize a two-channel optical programmable beamformer. The system is designed to achieve four-bit resolution. The architecture of the programmable dispersion matrix is based on an array of four delay-lines each having two spliced fiber Bragg gratings. We have experimentally investigated the optical signal processing performance of the optical beamformer in receive and transmit modes. Beampatterns for RF range 40 - 200 MHz are presented along with the theoretical calculations. The main lobe of the beampattern is shown to be independent of frequency for several target positions thus demonstrating a “squint-free” characteristic of this optical processor.

We propose and demonstrate a novel architecture for a programmable dispersion matrix (PDM) able to process two simultaneous independent RF beams with a resolution of 3 bits. The processor takes advantage of wavelength division multiplexing (WDM) encoding. In our two-channel approach, four adjacent ITU optical WDM channels are used as optical carriers. The odd and even optical carriers are independently modulated with each received RF-beam using electro-optic modulators. Thus, each pair of optical carriers directs RF information of one beam. The PDM is programmed to introduce independent true-time delays for even and odd optical channels. The PDM is based on three optical delay lines, arranged in a binary configuration. Each delay line consists of four fiber-Bragg gratings (FBG), with each FBG having a center-wavelength at one of the optical WDM channels. Beampatterns are characterized in receive mode for RF signals in 0.5-1.5 GHz frequency range obtained for target angles of 0° to 70°. The measurement results demonstrate the simultaneous processing of two received RF-beams and the squint free nature of the beamformer.

Differential mode delay is the main problem, limiting maximum length specifications for Gigabit Ethernet over multimode fiber optic cabling. A model of pulse propagation along multimode fiber, taking into account DMD, is proposed. Method of installed multimode fiber optic transmission lines examination, based on proposed model, is represented. This method had been successfully used for adaptation of combined MAN multimode fiber optic transmission line for Gigabit Ethernet, described for example in this paper.

A new loop output buffer architecture for optical packet switching is proposed. It is consisted of multiplex FDL loops, which are divided into k stages by k (M+1) x (M+1) switches. Using this architecture, a few FDL loops can provide large optical memories and get good packet-loss performance, and the delay performance can be improved by adding switches and buffer scheduling.

In this paper, we analyze the drawback of tail-dropping contention resolution in optical burst switched networks. Once contention occurs, we adopt modified head-dropping policy to resolve contention. This policy drops the head of the contending burst only if the overlapping region of the two bursts is less than the whole contending burst size, otherwise drops the whole contending burst. In order to have a better support of differentiated service, a new burst assembly policy, namely, Priority-based Proportional Mixed Burst Assembly, is proposed. Simulation results show that the proposed scheme performs very well in terms of performance metrics such as the times of contention and packet loss probability.

This paper studies the problem of virtual topology reconfiguration for changing traffic patterns in wavelength-routed network. We formulate it as an optimization problem, which tries to find out the virtual topology with least lightpath changes with network performance within given bound, when traffic demand imposed on wavelength-routed network changes. Average Packet Hop Count (APHD) is used to measure the degree of network performance. A linear formulation is presented and solved by MILP (Mixed Integer Linear Programming) method. Then we analyze the relationship between lightpath changes and degradation degree of performance. It can be seen that a large cost reduction for reconfiguration can be achieved by giving up a small amount of network performance.

In this paper, the dynamic distributed traffic-grooming problem in WDM optical mesh network is investigated. We give the proposal of the number optimization of wavelengths and ports are evaluated in a class of IP/WDM distributed nteworks. We presented and compared the performances of dynamic distributed traffic grooming in regular and irregular optical mesh networks, and give the optimized number of wavelengths and ports in different network condition.

Photonic delay lines (PDLs) are a powerful and important technology for the implementation of wideband phased array antenna controllers since they solve many of the limitations of electronic controllers, such as large instantaneous bandwidth and tunable signal processing bandwidth, as well as immunity to electromagnetic interference and electromagnetic pulses. In this paper, we propose PDL architectures that are based on commercially available wavelength division multiplexing (WDM) network elements and the OptiFlex technology. Wavelength selective elements have been proposed and used in optical networks. They can operate simultaneously and independently on any of the incoming signals, that can consist of many different wavelengths. Accurate control of the time delays is an important issue in microwave photonics applications. A unique technology that can adequately address this requirement is the OptiFlex technology by OFS Fitel. The concept of the OptiFlex technology was invented and developed to form arbitrarily complex fiber-optic fabrics capable of transporting optical signals between termination points in network equipment with almost negligible loss. The integration of OptiFlex and wavelength selective network elements can provide novel photonic delay line architectures that can provide small size, low power consumption, and ease in assembly, providing at the same time very good RF performance.

Resonant enhancement of electro-optic modulators has been shown to greatly reduce the required switching voltage (V_π) of these devices for narrow-band applications. Standing-wave resonantly enhanced modulators have been widely explored in the literature, but suffer from finite transit time effects and constraints on the active region length. Both of these effects limit their frequency of operation. A novel traveling-wave resonantly enhanced modulator is presented here that eliminates both of these problems. An implementation of the enhanced modulator using loss compensation is shown to reduce V_π by a factor of 17.8, and reduce link loss in an analog radio-on-fiber link by 25 dB.

The demand for residential broadband connectivity is growing world-wide, spurring on research efforts to solve the 'last mile' problem. Radio-on-fiber (ROF) based networks are an attractive solution as they offer significant infrastructure cost savings and economically leverage existing optical infrastructure. This network topology would necessarily operate in the 30GHz, LMCS band (Local Multi-User Communication system) by virtue of its broadband requirements. Such high operating frequencies and bandwidths demand expensive RF hardware. Pure harmonic upconversion of the radio subcarrier is an attractive method of reducing the frequency requirements of the system oscillator, mixer, and optical modulator with minimal complexity, thus reducing infrastructure costs. Unfortunately, severe phase and amplitude distortion results from the harmonic upconversion process and is further compounded by chromatic dispersion. This paper characterizes these distortions, and introduces a mitigating phase and amplitude predistortion scheme for harmonic upconversion with chromatic dispersion in Mach-Zehnder/direct detection based optical links. As a proof of concept, a prototype Mach-Zehnder based, 3rd order harmonically upconverted 10.41GHz, 16QAM subcarrier radio-on-fiber system is described, and preliminary experimental results presented.

A novel all-optical packet switching is presented. Adopting optical-controlled optical switch and multi-wavelength optical label technology, all the extraction of optical packet header, recognizing of route information and control of optical switch matrix can be processed in optical domain without O-E/E-O conversion. In this switching, the header label pulses are separated to different optical paths to control the optical-controlled optical switch matrix. According to different sequence and different wavelength of the header pulses, the different optical routes can be established, and a kind of all-optical packet switching is achieved.

This paper proposes a new 2x1 node receiver architecture for all-optical packet switched networks, whose basic principle can be extended to construct a larger buffer if more 2x2 switches and fiber delay lines (FDLs) are available. The most striking characteristic of the architecture is that the buffer is achieved by a combined use of feed-forward and feedback FDLs. We describe and analyze the concrete two-stage FDL architecture which is the cheapest and simplest multi-stage solution. As for control strategy, we demonstrate that bottom control strategy (BCS), which uses FDL that is arranged in the feed-forward form when both feed-forward and feedback FDLs are available, is a more robust solution against bursty and non-bursty traffic. Simulation results show that the developed scheme (1) performs well in terms of performance metrics such as packet loss probability; (2) has a simpler control requirement and a better packet loss probability than M-Quadro architecture.

This work proposes and analyses a sparse reconfigurable optical add/drop multiplexer consisting of eight tuneable ring resonator-based optical filters, and two eight channel arrayed waveguide routers. The design features a 0.25 add-drop factor, 32 channels in a four-skip-one architecture at 0.8 nm channel spacing, covers the conventional band, and takes advantage of the main limitation of single ring resonator-based optical filters, namely the rather limited free spectral range. Coupled mode theory and its transfer matrix formalism have been used to design the tuneable add-drop filters, while classical theory of signal spectral analysis have been used for determining its performance characteristics. The computed results for a sparse tuneable optical add-drop multiplexer equipped with single-ring add-drop filters were add/drop insertion loss 0.77 dB, through adjacent insertion loss 2.7dB, worst add/drop adjacent cross talk -43dB, add/drop bandwidth at full width half maximum 0.221 nm, and group delays between 0.05 ps and 2 ps for add/drop and through optical channels. The proposed tuneable OADM exhibits a very good magnitude response and properly adds, drops, removes and/or lets through the selected channel without disturbing the other channels.

Possible switching architectures, with Optically Programmable Logic Cells -- OPLCs -- will be reported in this paper. These basic units, previously employed by us for some other applications mainly in optical computing, will be employed as main elements to switch optical communications signals. The main aspect to be considered is that because the internal components of these cells have nonlinear behaviors, namely either pure bistable or SEED-like properties, several are the possibilities to be obtained. Moreover, because their properties are dependent, under certain condition, of the signal wavelength, they are apt to be employed in WDM systems and the final result will depend on the corresponding optical signal frequency. We will give special emphasis to the case where self-routing is achieved, namely to structures of the Batcher or Banyan type. In these cases, as it will be shown, there is the possibility to route any packet input to a certain direction according to its first bits. The number of possible outputs gives the number of bits needed to route signals. An advantage of this configuration is that a very versatile behavior may be allowed. The main one is the possibility to obtain configurations with different kinds of behavior, namely, Strictly Nonblocking, Wide-Sense Nonblocking or Rearrangeably Nonblocking as well as to eliminate switching conflicts at a certain intermediate stages.

We construct a multi-stage optical switch model with multilayer Kautz graphs. The switching path is built up by using the routing principle in the regular graph. A 3-stage switch is demonstrated. The modeling study shows pretty good statistics in terms of latency, packet loss and capacity. Implementation of this architecture with star couplers and other optical components is also discussed.