The design, upgrade and specification of optical communication systems depend on accurate computer models of predicted performance. NPL has been developing computational models of lightwave components and optical communication systems and has used its strong expertise in metrology to bridge the gap between measurement and simulation.

The computer models will be used to:

• understand the specification and measurement requirements of each component
• underpin the basic metrology by allowing the modelling of uncertainties
• assess existing installations for upgrade
• predict the performance of a communications link

Initial activity has been to model a basic 2.5 Gbit/s Metro type link. The link consisted of a directly modulated DFB laser transmitting over 108 km of standard single-mode optical fibre with an APD-FET receiver. Each component in the system was modelled separately and validated against experimental data. The real and theoretical components were then brought together to model the complete optical communication link and to provide real experimental validation.

Figure 1: Layout of the experimental and theoretical implemented optical communications link

The real and theoretical optical links were tested with a pseudo random bit stream and the bit error rates have been compared as a function of the decision threshold.

Figure 2: Simulated (solid blue) and measured (dotted red) signals after 108 km

F Figure 3: Simulated and measured error rates for a 108 km link

Parameter Extraction for the DFB Laser Model

A significant aspect of real system modelling is obtaining accurate model parameters for the real components used. In the case of the DFB laser, the lasing chip was modelled using a set of rate equations containing 7 parameters including the carrier lifetime, photon lifetime and gain. We obtained these parameters by measuring the relative intensity noise (RIN) of the laser output as a function of the drive current. The parameters were then extracted by least squares fitting the analytical expressions for RIN to the measured data. The figure below shows the measured RIN for three injection currents and the final modelled RIN of the laser output based on the parameters extracted by least squares. The use of a RIN measurement to provide data for the parameter extraction avoids the need for dynamic modulation that will be affected by laser packaging parasitics. This allows the laser chip and laser packaging parasitic models to be separated. A full description has been published in the Journal of Quantum Electronics, see reference I. Fatadin, D. Ives, and M. Wicks, "Numerical simulation of intensity and phase noise from extracted parameters for CW DFB lasers", IEEE J. Quantum Electron., vol. 42, no. 9, pp. 934 - 941, Sept. 2006.

Figure 4: Simulated (Solid lines) and measured (Points) RIN spectra for different laser injection currents

For further information, please contact: Irshaad Fatadin