Synthetic Electromagnetic Emissions: A New Approach to EMC Compliance Testing

We propose a novel data-driven approach for synthesizing unintended emission signals that can then be used to improve and optimize regulatory thresholds. The main component of our approach is a GAN's generator which is trained with a collection of unintended emission and then, in the inference phase, is asked to produce new and similar signals. We demonstrate that, with correlation as the measure of similarity, the GAN-generated signals are similar to the dataset.