AMS circuit modeling by Bayesian Model Fusion

The high data usage, scaling technology, and high simulation costs that are encountered when modeling the circuit performance of analog and mixed-signal circuits are some of the challenges that these models confront. A novel strategy is put forth to deal with this conflict, and a new modeling algorithm called Bayesian Model Fusion is used (BMF). The key element of BMF is the use of previously borrowed data to model the circuit’s performance at a later stage. The Bayesian Inference principle is used to extract data from the early stage. It makes use of Bayes theorem-based pre-silicon validation and post-silicon tuning methodologies to make this possible.