Metode de Machine Learning pentru suportul verificării circuitelor integrate analogice / Machine Learning methods for supporting verification of analog integrated circuits

Analog integrated circuit (IC) verification is a major component in the development of analog circuits because it ensures that the circuits will work as intended. It is also an essential inspection step because it helps identify and rectify any defects or errors introduced during the design process. In the current work, we propose a method to make the process of manual verification of signals more efficient. This is accomplished by the process of signal clustering, which enables the analysis of a vast volume of data without the need to individually examine each signal. The purpose of this clustering procedure is not to verify the requirements, but rather to visually inspect the signal in order to confirm the absence of unexpected oscillations, overshoots, undershoots, or other forms of glitches. The main objective of this method of optimization is to reduce the amount of time required to visually verify analog IC signals, while a second objective is to support IC fabrication by optimizing the process on how the production test sensor signals are analyzed. In this work we managed to improve the manual signal verification process both pre-silicon and post-silicon by using different implemented algorithms. We also managed to improve the analysis process of the sensors used in wafer production.