As usual in SMACD Conferences, the 2019 edition will also hold a collection of Special Sessions on relevant and emerging topics.
Today’s integrated circuits industry is characterized by complex systems-on-a-chip architectures where digital signal processing, analog interfaces, and memory blocks are integrated together on the same die. However, unlike digital circuits where the low-level phases of the design process are automated using established methodologies, the layout of analog and radio-frequency circuits is mostly drawn manually using computer-aided design frameworks. Although its automation has been intensively studied in the last three decades, still, there is no established automation flow in the industrial environment, resulting in a time-consuming and difficult-to-reuse design methodology.
Moreover, as analog and radio-frequency integrated circuit design moves to deeper nanometer technology nodes, the ever-increasing number of design rules and topological constraints further hamper the maturing of these tools. The aim of this Special Session is to bring together new research contributions dealing with analog and radio-frequency layout synthesis, spanning from the device methodologies up to the system level.
There are two buzzwords around IC world these days: Machine Learning (ML) and Deep Learning (DL). The truth beyond all the fuss is that, for example, machine / deep learning algorithms are already aiding anomaly detection on assembly lines in the industry. Moreover, in the last years, EDA vendors have improved their ML know-how, and, ML research projects are emerging everywhere. They promise to reduce design costs and expensive product recalls dramatically. Once that happens, we’ll reach the Golden Age of machine learning in EDA, of course, digital EDA… How about analog and mixed-signal EDA?
Recent works show Deep Neural Networks that can not only replace computationally demanding simulations (EM Simulation, Circuit Simulation or Parasitic Estimation) but also to address other aspects of Analog IC design, such as device sizing, device placement, or, circuit partitioning. This Special Session aims to bring together original research that advances analog and mixed-signal EDA using machine learning methods.