This paper reports on the novelty of the bacteriorhodopsin photonic crossbar system for broadband communications. A free- space dynamically reconfigurable bacteriorhodopsin holographic crossbar provides a large interconnectivity density, transparent data redistribution, and fiber optic bandwidth capacity. This switching device resolves optical-to-electronic and electronic-to-optical conversion bottleneck and reduces signal-to-noise degradation which is due to the conversions. This crossbar design is completely free of internal blocking which is one of the major drawbacks of guided optical crossbars. This system takes advantage of the parallelism and multidimensionality inherent in optics and can be scaled to a large capacity of NXN while it maintains a low weight and portability which are a projected requirement for future broadband communications. This NXN crossbar is an intelligent holographic system for routing and switching by dynamically reconfigurable grating of bacteriorhodopsin which has high write/read photocyclicity that is greater than 10⁶.

The optical N-gon prism switch is the simplest multi-layer architecture and a starting point for novel all-optical switching architectures. The N-gon prism switch is composed of parallel waveguides at each layer; switching between the waveguides at each layer and switching between the waveguides at adjacent layers is assumed, and the optics comes in by the transmission of light through the waveguides. Throughout the paper, switch setting algorithms for the N-gon prism switch are discussed where (1) an approach to an algebraic and scalable algorithm is presented for its fast implementation and (2) the further development of the multi-layer switching concept is taken into account.

Laser phase noise conversion to intensity noise due to fiber dispersion may impose severe degradation in intensity modulation and direct detection (IM-DD) system performance. In this paper, we investigate theoretically and numerically the influence of fiber nonlinearity on the conversion of phase noise to intensity noise at fiber output. Excellent agreement of noise characteristics between simulations results obtained from an accurate solution of the noise driven laser rate equations and theoretical predictions has been achieved. Results reveal that due to the fiber nonlinearity a significant enhancement of magnitude of the relative intensity noise (RIN) spectrum as well as its broadening is observed in the anomalous dispersion regime while a magnitude reduction and narrowing of the spectrum is observed in the normal regime. The system penalty due to RIN at the fiber output in IM-DD transmission systems using external modulator is assessed. The maximum transmission distance imposed by RIN for 1 dB of power penalty is presented for various optical power at fiber input in gigabit-per-second transmission systems. It is shown that in the anomalous dispersion regime the RIN enhancement due to the fiber nonlinearity can impose a significant impairment in IM-DD transmission systems through the phase modulation to IM conversion enhancement.

Measurements techniques relying on the intensity modulation (IM) fiber transfer function have recently been presented for estimating the fiber dispersion coefficient as well as the transient and adiabatic laser chirp. A theory which quantifies accurately the interaction of fiber nonlinearity and dispersion and their effects on the fiber transfer function is highly desirable to design properly long-haul gigabit-per- second transmission systems. In this paper, we investigate theoretically and numerically the influence of fiber nonlinearity on the IM fiber transfer function. A new small- signal theory for the optical fiber considering the fiber nonlinearity, dispersion and loss is presented. From this, the fiber transfer function with either an external modulator or a directly modulated singlemode laser as optical transmitter, has been computed for various optical powers. All theoretical predictions have been compared with simulation results and excellent agreement has been achieved. The results reveal that, in comparison with linear transmission, the fiber transfer function is enhanced by several dB in case of propagation in the anomalous dispersion regime and is reduced in the normal dispersion regime, even for moderate optical power at fiber input (approximately equals 20 mW). This means an increase of bandwidth in the anomalous regime and a decrease in the normal regime.

The main objective of this paper is to investigate the application of an erbium doped fiber amplifier (EDFA) in a coherent receiver using the pulse position modulation (PPM) format. Such a combination can exploit the abundant monomode fiber bandwidth to offer improved sensitivity and improved multichannel selectivity through the exploitation of coherent transmission. Original theoretical results are presented for a system operating in thermal noise dominant condition at a bit rate of 622 Mbit/s and a wavelength of 1.53 micrometer. By using a suboptimal matched filter the results demonstrate that the system offers 7.5 dB sensitivity improvement over a homodyne PPM system and 9.4 dB improvement over PSK homodyne. The system investigated also offers a 10.4 dB sensitivity improvement over an equivalent optically preamplified direct detection (DD) PPM system. It is also demonstrated that by using an optical filter in order to reduce the ASE noise due to EDFA the improvement in receiver sensitivity was only 0.3 dB. This greatly reduces the requirement for optical filters in coherent systems and therefore the full bandwidth of the optical amplifiers can be utilized.

A new hybrid optical wireless system based on pulse position modulation (PPM) and PIN-BJT receivers is proposed and its performance is evaluated. The system is operated at a bit rate of 10 Mbit/s and (lambda) equals 0.85 micrometer. The receiver achieves a low noise current of about 2.7 pA/(root)Hz. Original results are presented demonstrating that the system can achieve a sensitivity of -50.5 dBm at an error rate of 10^-9. This represents a sensitivity improvement of 9.8 dB over a comparable ON OFF keying (OOK) PIN-BJT system. This paper also shows that shot noise, from background light, can reduce sensitivity by about 10.2 dB - 5.4 dB when the background light power is around 5 mW - 0.5 mW and hence careful consideration has been given to ambient noise. The overall sensitivity enhancement has significant impact on the optical wireless network, where the maximum allowed transmitted power is limited by safety issues and background noise. Moreover, the use of PPM offers other advantages such as the low average power which is suited for battery operation and a high peak power that aids detection.

An optical backbone network based on WDM (or OTDM) technology may become an economical choice for providing future broadband services. To achieve a balance between the coarse-grain optical circuit switching (via wavelength routing) and fine- grain optical packet/cell switching, optical burst switching is proposed. We study a one-way reservation protocol called just-enough-time (JET), which is suitable for switching bursty traffic in a high speed optical backbone network. The JET protocol has two unique, integrated features, namely, the use of delayed reservation (DR) and buffered burst multiplexers (BBM). By virtue of DR, the JET protocol not only increases the bandwidth utilization, but also facilitates intelligent buffer management in BBMs, and consequently results in a high through-put. Both analysis and simulation results show that the JET protocol can significantly outperform other one-way reservation protocols lacking one or both of these features.

By means of numerical simulation, we investigate the implementation of duobinary coding at 20 Gbit/s transmission over standard singlemode fiber and compare its performance with the performance of binary coding. Two system configurations are compared: duobinary coding implemented at the transmitter (DIT configuration) and duobinary coding implemented at the receiver (DIR configuration). Numerical results reveal that duobinary coding is much less impaired by fiber transmission than binary coding. However, the binary back-to-back receiver sensitivity can be as much as 9 dB better than the DIT configuration if a high-gain preamplifier is considered. The DIR configuration achieves an improvement of about 2 - 3 dB in the back-to-back receiver sensitivity because of the smaller noise bandwidth. It is shown that the DIT and DIR configurations exhibit similar dispersion penalty. Nevertheless, the DIR configuration achieves an improvement of about 2 - 3 dB in the back-to-back sensitivity because of the smaller receiver noise bandwidth.

An architectural approach for indoor optical wireless multimedia networks is proposed. Optical wireless (OW) systems offer user mobility as is the case with rf wireless. However, OW offer additional advantages such as increased bandwidth, abundant unregulated spectrum, freedom from fading and a radiation that is contained within the communication. Zone in this paper the rationale behind the approach, the important role of the wireless packet transfer layer (WPTL) in the protocol stack of the system, and the transport architecture of the ATM cells through the optical wireless channel are discussed.

In this paper, we introduce a structural model for WDM (wavelength division multiplexing)-based LT (line termination) systems containing the fundamental functions recommended by ITU-T, and we describe how to configure the supervisory system and data communication channels to be adaptable to the WDM- based transmission systems considering compatibility with the SDH-based supervisory systems. Also we propose that two new reference points (N2 and N3) be added to the MCF (message communication function) like N and P from the view point of the telecommunication management network.

The managed all-optical WDM transport network with the all- optical cross-connects is offering not only the high bandwidth, which can be shared between arbitrary clients accessing the network, by static allocation of the trails (also denoted as optical paths) or the virtual sub-networks, but also the service to perform reconfiguration of such virtual sub-networks and to reassign the capacity according to the demands and services required from the client, by the management actions. To fulfill the management task of providing the optical path, which is characterized by a set of optical parameters in the optical domain, the optical network management system must act upon the requirements of the different clients. This imposes that these requirements have their representation in the optical domain and vice versa. In this article we are investigating the possible approaches to WDM network management, first starting from the assumption that the capacity in the network can be arbitrarily shared between different transparent optical paths, and then introducing the analogous transmission parameters related to the transmission equipment and media which restrict this capability and make the management of the optical path dependent on the different types of the clients and different utilization of the capacity in the network. It is shown how those parameters are to be assigned to the different layers in the proposed layered architecture model of the optical network. This model is also discussed from the view point of the management integration and the optical QoS parameters mapping to the client QoS parameters are proposed.

The deployment of an all optical network requires that all transmission and switching equipment be properly synchronized. The synchronization performance characterization of telecommunication equipment and networks requires a wide variety of specialized equipment, each one performing a specific task, with limited data collection, transfer, and analysis capabilities. Most of the required equipment is heavy, difficult to use, and occupy substantial floor space. A portable, user friendly, and easy to maintain automatic test station for synchronization performance monitoring and characterization has been developed.

Increasing communication speed requirements have created a great interest in very high speed optical and all-optical networks and interconnects. The design of these optical systems is a highly complex task, requiring the simultaneous optimization of various parts of the system, ranging from optical components' characteristics to access protocol techniques. Currently there are no computer aided design (CAD) tools on the market to support the interrelated design of all parts of optical communication systems, thus the designer has to rely on costly and time consuming testbed evaluations. The objective of the CATO (CAD tool for optical networks and interconnects) project is to develop a prototype of an intelligent CAD tool for the specification, design, simulation and optimization of optical communication networks. CATO allows the user to build an abstract, possible incomplete, model of the system, and determine its expected performance. Based on design constraints provided by the user, CATO will automatically complete an optimum design, using mathematical programming techniques, intelligent search methods and artificial intelligence (AI). Initial design and testing of a CATO prototype (CATO-1) has been completed recently. The objective was to prove the feasibility of combining AI techniques, simulation techniques, an optical device library and a graphical user interface into a flexible CAD tool for obtaining optimal communication network designs in terms of system cost and performance. CATO-1 is an experimental tool for designing packet-switching wavelength division multiplexing all-optical communication systems using a LAN/MAN ring topology as the underlying network. The two specific AI algorithms incorporated are simulated annealing and a genetic algorithm. CATO-1 finds the optimal number of transceivers for each network node, using an objective function that includes the cost of the devices and the overall system performance.

All-optical networks can be dynamically reconfigured to adapt to changing traffic patterns. Many researchers, however, concentrate on reconfiguration of opto-electrical networks, as it is more practical at the present level of technology. This work presents an ad hoc algorithm that calculates transition from an existing configuration of an all-optical network to another, with minimal disturbance of the established connections. It analyzes the algorithms' wavelength usage efficiency and long-term stability, showing that such or similar algorithms can be used for practical purposes until some strict theoretical approaches can be found.

A wavelength division multiplexed (WDM) broadcast star network is considered. The network has N tunable transmitters and receivers with nonnegligible tuning delay T. The network traffic is composed of continuous flows whose transmissions may be preempted and resumed. The traffic is delay constrained so that a bit cannot be delayed by more than a predefined number of time units denoted by D. Guaranteed nodal throughput results are presented for traffic patterns that are fair. In particular, if the nodal throughput is (lambda) then all nodes can transmit/receive at rate (lambda) as long as they do not transmit/receive more than this rate. When T very much less than D and NT very much greater than D, then the upper and lower bounds for (lambda) are approximately 4 D/NT and 1/4 D/NT, respectively. Notice that the conditions T very much less than D and NT very much greater than D mean that a single tuning delay is a small contribution to delay, but the aggregation of all the tuning delays may be a large contribution to delay.

This paper describes how an experimental full services access network has been constructed at BT Labs and presents views on how its performance could be improved to meet the reliability and traffic loading requirements expected in real applications such as fiber to the business and fiber to the cabinet. The experimental network included: asynchronous transfer mode (ATM) switch, an ATM passive optical network (PON), very high speed digital subscriber loop (VDSL) customer drop and ATM forum 25 Mbit/s customer network. The design and realization of the VDSL customer drop, the signaling system and the interfaces between the system elements formed a major part of the design and construction work at BT Labs. The ability to cope with varying service demand and achieving the necessary quality of service are important requirements for roll-out systems. This paper describes how these requirements could be met in the design of future proprietary equipment.

The broadband loop project is a joint European project which is developing low cost access network solutions for competitive access providers. The project is developing passive optical network (PON) solutions optimized for delivery of services in low penetrated areas. The basic version of the PON is a SDH PON which can provide managed SDH transport to the customers premises. ATM can be provided via the SDH PON system. The PON bandwidth can be gracefully extended from 155 Mbit/s bidirectionally up to provide 1 Gbit/s in the downstream direction and 576 Mbit/s in the upstream direction. Subcarrier multiplexing is used to extend the bandwidth. Digital subscriber line (DSL) technology is used to provide broadband services over existing copper wires. The project is evaluating the tradeoff between use of optical fiber or copper in the access network. Life cycle cost studies compares different deployment scenarios for business and residential subscribers. Field trials are installed in Denmark, Portugal and Poland in order to evaluate the system under real life conditions.

This paper presents the use of distributed, intelligent control and management in optically amplified repeaters. These optical repeater units (ORUs) are used in an optical access network. A semiconductor optical amplifier (SOA) has been used in the upstream direction because of the possibility of fast switching. The real time control platform consists of both a hard- and a software part. The software control is handled with the embedded control system FORTRESS developed by IMEC.

Power budget dimensioning of semiconductor optical amplifier based SuperPON architectures originating from the EC ACTS project PLANET is described. The upstream power budget is critical for SuperPONs. Under worst case conditions it is found that a 5 SOA cascade in the upstream is cable of achieving the full PLANET target system dimensions (2048 way split and 100 km range). Longer feeders are also found to be possible, with the use of further feeder amplification stages, which is useful for the situation when the SuperPON is to be deployed in a ring trenching structure.

All-to-all personalized communication (AAPC) requires each of the N nodes in a network to send a unique message to each of the other N - 1 nodes. In this paper, we study the problem of scheduling AAPC in WDM rings with K wavelengths and T transmitter-receiver pairs per node. We first determine the theoretical lower bound on the schedule length in a unidirectional ring with negligible tuning delay. A scheduling method based on a 'recursive packing procedure' is then proposed. Two extensions of the above work are also made. First, we use a similar method to schedule AAPC in bidirectional rings with a small T, which achieves a better schedule performance than the method in reference one. Second, we modify the schedule to take into account the effects of non-negligible tuning delay. The schedules derived in this paper can achieve the minimum length for some values of K and T, and a near minimum length in other cases.

In this paper, we analyze the performance of a packet-switched broadcast-and-select all-optical network with N nodes and W wavelengths. We model the network as an N X N space- division time-slotted packet switch. We consider the effects of the input queueing dynamics on the network performance, and present a queueing analysis of networks supporting multicast traffic with fanout splitting. We obtain the maximum supportable throughput as a function of the network size, the number of wavelengths, and the multicast size. The average multicast packet delay is evaluated as a function of network and traffic parameters. It is found that increasing the number of wavelengths beyond 40% - 60% of the number of nodes results in little improvement in network performance. We also study a second switching scheme that yields higher overall network throughput at the expense of introducing packet dropping.

By providing several parallel high-capacity channels in the same fiber, WDM technology is commonly considered to be one of the fundamental solutions for future high speed computer networks. By means of WDM the potential THz bandwidth of the optical fiber can be multiplexed into channels with transmission speed compatible with opto/electronic and electro/optic converters -- e.g., transmitters and receivers. However, a fundamental scalability problem has to be solved in order to build larger than LAN size WDM packet switched all- optical networks. Current solutions can not be scaled by simply adding nodes to the existing system due to the limited number of wavelengths and/or the available power budget. The lack of modular solutions, similar to those used in traditional electronic networks, can be attributed to the unavailability of optical technology that can support switching of WDM channels on a per packet basis. Recently, a novel approach was proposed to obtain a packet switching WDM network which is scalable in coverage area, size and number of wavelengths. The approach is based on the combination of two ideas -- use of a backbone that connects (bridges) multiple LAN segments, where bridging is done using the new concept of photonic slot routing (PSR). In this concept, packets transmitted simultaneously on the WDM channels in any 'photonic' slot are switched jointly in the bridge between the LAN segments and the backbone. Thus packets in a given slot remain transmitted on the same path, from their source LAN to the destination LAN and consequently, the bridge is only required to handle complete slots, i.e., all WDM channels jointly. The aim of this work is to present the first analytical model developed to evaluate the performance of the photonic slot concept applied to a single fiber ring backbone system, independently on the interconnected LAN segment topologies, e.g., ring, folded bus, star.

Wavelength-division-multiplexing (WDM) techniques are now widely spread for high bit rate transmission. Beside the increase of the transmission capacity, WDM should be used in flexible, transparent and cost effective solution for routing large amount of traffic in the optical domain. WDM networking implies a novel network analysis in order to define the traffic transported and routed optically, the protection/restoration schemes more adapted to the optical domain and the reconfiguration possibilities. The WDM optical layer dimensioning is presented here and shows the design and optimization of this transport layer in terms of transmission resources (wavelengths, fibers, amplifiers) and optical node functionalities (optical add-drop multiplexers, optical cross- connect, wavelength translation, . . .). A dimensioning methodology is established, defining the main inputs of this task and the main issues. In this paper different options and issues of the WDM dimensioning are presented and discussed. First, the optical path routing is studied and different routing strategies are given for the minimization of specific parameters. The second issue concerns pure resource allocation. Two different approaches based respectively on wavelength routing and wavelength translation are introduced. The benefits of wavelength translation is discussed and different criteria for the location of this functionality are established. The discussion and analysis are illustrated on several test cases covering different network topology assumptions.

Wide area all-optical networks are emerging as practical and future-proof solution to the enormous bandwidth requirements of today's end-users. Over the past few years, the field of all-optical networks has experienced tremendous amount of research, development, and prototyping activities. Control and management issues of all-optical networks, which are getting increasing attention in the research and development communities, are considered in this paper. We present a new approach for modeling call blocking probability in wavelength- routed wide-area all-optical networks. When an optical connection (lightpath) request arrives at a node, first a route and a wavelength along that route is selected to establish the requested connection. If an available route or wavelength doesn't exist then the connection is blocked at the source node. If a route and an available wavelength along that route are found then the wavelength along the different links of the route are reserved as soon as possible. However, reserving the selected wavelength on all the links of the route takes finite amount of time which is dependent on the propagation delays of the links and processing speeds of the nodes along the route. Hence, the selected wavelength on one or more of the links along the route might be reserved by another lightpath request while the set-up procedure of the previous one is in progress. One of the primary goals of this work is to study blocking of lightpath requests at intermediate nodes. Note that, although several researchers have studied blocking at source node in optical networks, to our knowledge blocking characteristic at intermediate nodes has not been studied earlier. A new lightpath set-up approach is also proposed. Simulation results show that the proposed approach can reduce blocking at intermediate nodes. Numerical results from analytical as well as simulation models are presented for various network configurations.

In this paper, we consider the problem of optimally placing a given number of wavelength converters on a path to minimize the call blocking probability. Using a simple performance model, we first prove that uniform spacing of converters is optimal for the end-to-end performance when link loads are uniform and independent. We then show that significant gains are achievable with optimal placement compared to random placement. For non-uniform link loads, we provide a dynamic programming algorithm for the optimal placement and compare the performance with random and uniform placement.

It is true that in all-optical networks, network performance can be improved by wavelength conversion. However, the switching node with wavelength conversion capability is still costly, and the number of such nodes should be limited in the network. In this paper, a performance optimization problem is treated in all-optical networks. We propose a heuristic algorithm to minimize an overall blocking probability by properly allocating a limited number of nodes with wavelength conversion capability. The routing strategy is also considered suitable to the case where the number of wavelength convertible nodes are limited. We validate the minimization level of our heuristic algorithm through numerical examples, and show that our algorithm can properly allocate nodes with conversion for performance optimization.

In this paper, we study an adaptive algorithm for routing in an all-optical wavelength routed network where a lightpath is dynamically created in response to a request for communication and the WDM channels constituting the lightpath are reclaimed when the communication is over. We have proposed three possible distributed strategies to determine, if possible, a lightpath when there is a request for communication. Each of these strategies have their advantages and disadvantages in terms of the expected blocking probability and the set-up time. We consider the Kautz digraph as the network topology and study the performance of our distributed strategies through simulation.

This paper presents some results of a SuperPON (lab) demonstrator. The SuperPON is an ATM based passive optical (access) network, intended to support a large split (greater than 2000), and a long range (80 km, of which 70 km in the feeder section, and 10 km in the drop). The overall network capacity is 2.5 Gbit/s TDM downstream at 1550 nm, and 155 Mbit/s TDMA upstream at 1310 nm. The first (lab) demonstrator has been built around an APON system, into which a 2.5 Gbit/s downstream upgrade has been introduced, together with optical amplification to overcome large network losses. This optical amplification consists of EDFAs in the downstream, and switched SOAs in the upstream direction. On the demonstrator set-up, various measurements have been performed to look for the acheivable power budget that can be tolerated in the different network sections. The demonstrator does not only show the feasibility of the architecture and optical technologies, but also demonstrates the capabilities of real services.

In reference one, a network architecture called hyper-cluster employing wavelength-routing, multi-hop packet switching and optical reconfiguration is presented. It is modularly scalable to very large configurations on both hardware and operational bases. A hyper-cluster uses a logical hierarchy for the purpose of addressing but guarantees that all access nodes have a constant number of transceivers. It is a cluster of regular graphs; the clustering structure follows the traffic distribution in a grand granularity. The issue of operational scalability is addressed by presenting a scalable routing protocol and a scalable reconfiguration protocol. When using shuffle-net as the building block, a novel routing scheme called quantified deflection routing is presented. The scheme improves call blocking performance significantly. In this paper, we propose a distributed reconfiguration protocol. The network throughput and virtual call blocking performance is obtained via simulation on large networks (with size beyond 200 nodes). Numerical results show that the dynamic self- routing protocol, combined with quantified deflection routing for shuffle-net, can achieve excellent resource utilization efficiency for very large networks. When the call arrival rate is below 0.3, the capacity provided by the hyper-cluster dynamic routing algorithm is close to that of an infinite capacity centralized switch (lowest possible call blocking caused exclusively by congestion on the finite capacity user input/output links, never by the switch fabric itself).

All-optical communications based on OFDM techniques are very attractive for future telecommunications. However, the frequency instability of semiconductor laser diodes still constitutes an obstacle to exploit OFDM benefits. In this paper, we present a decentralized frequency control system suitable for large scale all-optical networks. Spectrum partitioning is proposed so that the controllable spectral range could be enlarged and link fault could be detected. Experimental results on monitoring and stabilizing channel frequency in a local node, confirm the feasibility of the described method.

A new layered structure for multi-wavelength photonic transport networks is proposed after optical wavelength- division-multiplexing (WDM) technologies and optical path technologies are introduced. Photonic layer is inserted between traditional physical layer and section layer. It can be divided further into three sublayers: optical path layer, WDM section layer, and multi-wavelength amplifying section layer. The functions of photonic layer and its sublayer are defined. Some problems that are related to photonic layer are discussed and solved. Network management issues are also discussed in this paper.