Autonomous Data-driven Model for Extraction of VCSEL Circuit-level Parameters

In recent years, a number of computationally efficient models have been developed that adequately describe the static and dynamic behavior of the Vertical Cavity Surface Emitting Laser (VCSEL). In order to correctly recreate the behavior of existing laser sources, a large number of physical parameters must be specified. Finding these unknown physical characteristics in experimental curves may be time-consuming, and mainly requires trial and error processes or regression analysis. Instead of manually analyzing experimental data to find the best VCSEL parameters, we propose a Machine Learning (ML) based solution to automate the process. The proposed approach exploits the parametric dataset obtained from Light-current and Small-signal modulation responses to extract the required model parameters. Excellent results are obtained in terms of relative prediction error.