In frames of EURESCOM 615 project several optical network architectures have been analysied from techno-economical point of view. This paper presents results from availability analysis of two optical rings. Need for an efficient combination of optical protection on architectural level and HW protection on electrical equipment level is identified. Improvement of the availability of an optically protected self-healing ring architecture by electrical equipment HW protection is presented and analysed.

We report the theory of operation and time domain model (TDM) simulation results for a two-section Fabry-Pe´ rot sampled grating distributed Bragg reflector (FP/SGDBR), 100-GHz, step-tunable laser [sidemode suppression rate (SMSR)>38 dB]. Although the main idea, the Vernier effect, was previously published, we present for the first time these two-section combinations for this specific sort of application.

Simulation studies of dynamic routing with rearrangement in WDM ring networks indicate that up to five addition paths maybe accommodated with 16 wavelengths. Dynamic alteration of a wavelength is necessary to achieve this.

A dual mode laser source is proposed. It is based on a sampled grating DFB laser constituted with alternating DFB and FP sections. The basic principles of this device are presented, and its behaviour is studied through modelling. It is shown that mode spacing in the millimetre-wave domain (here 60 GHz) can be achieved with relatively short devices using a common DFB structures coupling coefficient

A heuristic methodology is proposed for the setting up of a stack of wavelength-division-multiplexing (WDM) rings with wavelength reuse when the design traffic exceeds the capacity of a single ring and no wavelength conversion is employed. A ring stack consists of an overlay of rings routed over the same physical route, and it can be setup and dimensioned in a myriad of ways. The design traffic comprises of a set of bidirectional lightpaths or wavelength connections. There exists a tradeoff between the number of nodes and the number of rings required to carry this traffic, and it is demonstrated that both cannot be minimized simultaneously. For certain traffic patterns, we identify stacks requiring the minimum number of nodes or WADM's, which is desirable from a cost point of view, and stacks requiring the minimum number of rings. An algorithm is presented that manipulates the tradeoff phenomenon to produce a spectrum of designs with deterministic composition. We finally conclude by identifying factors that may influence the choice of design.

The ACTS project HIGHWAY (AC067) addresses promising ultra-high speed optoelectronic components and system technologies for 40 Gbit/s time-division-multiplexed (TDM) transport systems. Advanced 40 Gbit/s TDM system lab demonstrators are to be realized and tested over installed field fiber testbeds. This paper reviews the current status of 40 Gbit/s TDM components and subsystem technologies achieved in HIGHWAY. The results of HIGHWAY 40 Gbit/s TDM systems and field tests will be reported in a subsequent paper.

In this paper, we present a new semiconductor laser design used for dual mode signal generation. This new design is based on sampled grating structures which have been mainly used, up to now, for tuneable laser applications. We will report the main theoretical design aspects of that structure and present simulation results using the Time Domain Model (TDM). We will also spot on the coupling coefficient effect on the mode spacing and point out an optimal coupling coefficient value. The elected structure is single section, single electrode and needs only one DC current

Channel provisioning in four-fiber wavelength-division multiplexing (WDM) rings is studied for realistic telecom-operator traffic. For static traffic, wavelength conversion is of no benefit. For dynamic traffic, especially with churn, 1+1 protected WDM rings may, surprisingly offer higher capacities. For both cases we present the most suitable wavelength assignment algorithms.

In this paper we propose a heuristic algorithm aiming to give a near optimal wavelength allocation, assuming no wavelength conversion (each demand is on the same wavelength from end to end) and no particular topologies. Moreover the demand routing is not the main topic of this paper and then is only quickly presented, even if this task strongly influences the results. We show that the wavelength conversion, with its additional complexity and its expense, does not provide a real advantage, in a ring network, compared to the no-conversion case with the heuristic. In addition, in meshed topologies, we show that this heuristic achieves good results without the time penalty and complexity expense due to a simulated annealing based algorithms.

Channel provisioning in 4-fibre WDM rings is studied for realistic telecom-operator traffic. For static traffic, wavelength conversion is of no benefit. For dynamic traffic, especially with churn, 1+1 protected WDM rings may, surprisingly, offer higher capacities. For both cases we present the most suitable wavelength assignment algorithms.

It is shown that in a four-fibre optical WDM ring with N nodes, the number of wavelengths required to accommodate a full mesh interconnection is (N-2 -1)/8, rounded up to the nearest integer. This result applies for all values of N, is the same with and without wavelength conversion, and represents the optimum value.

An 80Gbit/s (8 x 10Gbit/s) OTDM system, constructed entirely with electroabsorption modulators (EAM) as the active components, is presented. A tandem EAM pulse source with a simple dual frequency drive generates transform limited 4.5ps pulses. At the receiver, the 10Gbit/s channels are demultiplexed from 80Gbit/s in a single step using a low polarisation sensitivity EAM. Error free operation with a sensitivity of -21.5dBm is achieved.

40 Gbit/s OTDM transmission over 103 km of standard fibre, is demonstrated for the first time using polarisation-insensitive mid-span spectral inversion. The phase-conjugation is achieved by four-wave mixing in an optimised SOA, using two orthogonally-polarised pumps for polarisation-independence.

Nonlinearities in semiconductor optical amplifiers have been used to demonstrate a wide range of functions applicable to future optical networks such as wavelength conversion and optical switching. Four-wave-mixing effects in SOAs have been studied extensively in many laboratories with respect to the underlying physical processes and system applications. At BT Labs an optimisation of SOAs for FWM has been achieved by altering the device active layer composition and by increasing the device length. We will review recent progress at BT Labs in dispersion compensation, wavelength conversion and demultiplexing at bit-rates of 40 Gbit/s using these devices

A stacked ring design was advocated for the designing economical WDM ring networks and an algorithm was presented which confirmed the possibility of the cost-effective dimensioning of ring networks to meet demand growth. Two variants of the stacked ring design heuristic was described with Algorithm B yielding those designs requiring the fewest nodes. The described scheme can also be adapted for the design of shared-protection SDH rings with time-slot interchange (TSI) capability. In such rings, a load-balancing scheme can be employed to fully utilise the available bandwidth of the ring. Because each ring in the stack can vary in size, simple node/ring additions should be possible enabling the network to be cost-effectively augmented to meet further growth in demands. These heuristics offer simple real world implementations and their fast execution times lend to their implementation as frequently called routines in network design tools

In this article the increasing use of WDM systems in telecommunications networks is highlighted, and the potential role for wavelength conversion in future upgrades of such systems is reviewed. Techniques for achieving wavelength conversion by all-optical means using nonlinearities in semiconductor optical amplifiers are explained, and experimental results obtained at BT Laboratories are used to illustrate the state of the art.

An analytical theory of the dynamical behavior of erbium-doped amplifier (EDFA) chains is developed. A simple and detailed physical picture is presented, explaining the speeding up and the appearance of overshoots in the transient response. The speed of the transients is shown to be proportional to the number of the amplifiers in the chain.