Machine Learning based Performance Prediction of Automotive Microcontrollers

The automotive industry requires high quality and high safety for its products. During the manufacturing process, microcontrollers are subject to a performance screening to detect the under-performing devices. In recent microcontrollers, ring oscillators are embedded to measure several physical parameters. The main purpose of this work is the analysis of these measurements, called Speed Monitors, and consequently the development of a Machine Learning model to predict the maximum operating speed of the devices. The main challenges are the limited resources of microcontrollers and the restricted prediction time in the manufacturing process. The final goal is the definition of a screening methodology for the production test able to meet the requirements of the automotive sector.