دانلود کتاب طراحی مدار مجتمع آنالوگ تحت شرایط PVT: تکنیک‌های کارآمد تقویت و یادگیری انتقالی

The evolution of the electronics industry and the increasing demand for consumer products have impacted on the design of integrated circuits (ICs). Deep nanometer technologies allowed an increased complexity of ICs, providing smaller and more power-efficient solutions, but consequently became more challenging to design. The IC market has seen substantial growth, valued at 681.05 billion USD in 2024, and is projected to grow to 2062.59 billion USD in 2032. This growth underscores the importance of addressing design challenges in the IC industry, which impacts various sectors like education, health care, and transportation. The development of these systems is crucial to meet society’s evolving needs, driven by advancements in global communications and technology integration. When it comes to designing these systems, digital circuits are usually chosen to implement most of the functionalities due to the robust nature of handling digital signals, easier to reuse, and having many automated design tools within its design flow. However, most systems rely on interaction with the real world, and that is where analog, radio-frequency and millimeter wavelength (mmWave) ICs come into play. Systems where these two types of circuits are employed are denominated mixed-signal systems-on-a-chip, and can be found everywhere. In the past years, automatic simulation-based sizing approaches became essential in designing these analog, radio-frequency and mmWave IC blocks for modern applications to ensure their robustness. However, optimizations considering process, voltage, and temperature (PVT) corners or layout still pose unprecedented challenges in applying these tools due to the high simulation times and different simulator convergence issues. Therefore, the work presented in this book addresses the automatic sizing of analog ICs assisted by machine and deep learning. The first part of this work presents a comparative study of five different approaches for integrating PVT conditions, along with several reward functions, within state-of-the-art reinforcement learning-driven sizing methodologies. Results reveal the trade-offs of each approach, and the best-performing one for this problem is highlighted and discussed to facilitate more research activities within this field. Furthermore, considering these PVT corners during traditional simulation-based techniques is also incurring prohibitive optimization times. Therefore, in the second part of this work, an additional tech-nique is explored, where deep learning is used to assist a simulation-based sizing tool, via independent and parallel PVT performance regressors that bypass the simu-lator. These performance regressors are shallow artificial neural networks, where transfer learning from nominal to corner conditions was employed to avoid the need to acquire an expensive and time-consuming PVT training dataset, and an online Bayesian-assisted incremental learning is used to refine each model with accurate simulator data. This methodology is vastly tested on the design of millimeter wave-length ICs. When compared with full simulation-based synthesis, it reduced the workload of the circuit simulator up to 69% while achieving a speed-up factor of 3.4 × and requiring less 86% dataset generation effort when compared with most recent PVT regressors. Finally, the authors would like to express gratitude for the financial support that made this work possible. The work developed in this book was supported by national funds through FCT—Fundação para a Ciência e a Tecnologia, I.P., and, when eligible, co-funded by EU funds under project/support UID/50008/2025—Instituto de Telecomunicações, with DOI identifier 10.54499/UID/50008/2025, and also, by project ACTON, withDOIidentifier10.54499/2023.11981.PEX. This book is organized into four chapters. Chapter 1 presents an introduction to the analog, radio-frequency and mmWave IC design area and discusses how the advances in machine and deep learning can enhance existing electronic design automation tools in the analog spectrum. Chapter 2 presents a comprehensive study of the available tools for analog design automation. Focusing on most recent works where machine learning techniques are applied to analog IC sizing. Chapter 3 studies the incorporation of PVT corner conditions into reinforce-ment learning-based analog IC sizing, as it remains a challenging task. Therefore, a comparative study of five different approaches for integrating PVT conditions is conducted, where each approach begins with the same conditions to ensure a fair comparison, and is evaluated within the same circuit topology, comparing agent steps, number of simulations, execution time and final sizing functional behavior. Results reveal the trade-offs of each, and the best-performing one for this problem is highlighted and discussed to facilitate more research activities within this field. Chapter 4 considers PVT corners during traditional simulation-based synthesis of mmWave ICs in nanometer technologies, which is incurring prohibitive optimization times. An innovative research towards the automation of mmWave IC design by using deep learning to assist a simulation-based sizing tool, via independent and parallel PVT performance regressors that bypass the simulator, is presented. There, transfer learning from nominal to corner conditions avoids the need for expensive PVT data, and an online Bayesian-assisted incremental learning is used to refine each model with accurate simulator data.