Accelerating High-Frequency Circuit Design Using Advanced AI Algorithms

Historically, high-frequency circuit design for demanding applications has relied heavily on the expertise of seasoned circuit designers with decades of experience. This dependence on deep experience can hinder progress in a highly competitive environment, especially when the workforce is insufficient to meet the United States' commercial and defense electronic needs. At the same time, the demand for workforce development has grown, and design tools have become increasingly automated. The U.S. has a rich history of developing Electronic Design Automation (EDA) technologies, integrating advanced algorithms and cutting-edge technologies. In recent years, the scale and operating speeds/frequencies of integrated circuits (ICs) have increased exponentially, straining the scalability and reliability of traditional circuit design workflows. As a result, EDA algorithms and software must evolve to become more effective and efficient, particularly in managing high-frequency and high-speed applications with extensive parametric search spaces and minimal latency. The U.S. now faces two critical needs: (a) the immediate integration of AI algorithms to produce highquality, reliable, high-frequency circuit designs within significantly shorter design cycles, and (b) the rapid training of a skilled workforce capable of generating innovative designs at a pace that ensures the U.S. maintains its competitive edge. This presentation will discuss the specific use of AI algorithms in RFIC design. It will provide an assessment of the advantages and the complexities of this approach vis.a.vis the promise of a robust and autonomous design methodology that can easily be implemented on EDA platforms.