This item is under embargo and not available online per the author's request. For access information, please visit http://libanswers.wustl.edu/faq/5640.

ORCID

http://orcid.org/0000-0001-7539-8250

Date of Award

Summer 8-15-2021

Author's School

McKelvey School of Engineering

Author's Department

Electrical & Systems Engineering

Degree Name

Doctor of Philosophy (PhD)

Degree Type

Dissertation

Abstract

Analog/mixed-signal (AMS) integrated circuits (ICs) play an essential role in electronic systems by processing analog signals and performing data conversion to bridge the analog physical world and our digital information world.Their ubiquitousness powers diverse applications ranging from smart devices and autonomous cars to crucial infrastructures. Despite such critical importance, conventional design strategies of AMS circuits still follow an expensive and time-consuming manual process and are unable to meet the exponentially-growing productivity demands from industry and satisfy the rapidly-changing design specifications from many emerging applications. Design automation of AMS IC is thus the key to tackling these challenges and has been a longstanding open research question.

Towards the goal, this dissertation explores various machine learning (ML) techniques to automate AMS circuit design and infuse learning abilities in these circuits to meet diverse application requirements. First, a novel design framework is presented to use neural approximation--a type of supervised learning (SL) method to learn analog-to-digital converters (ADCs), one of the most essential AMS circuits, by training rather than design by hand. The method not only achieves an automated synthesis flow for various ADCs but also enables the designed ADCs of intelligence to flexibly learn different quantization schemes to accommodate miscellaneous applications. Our results demonstrate superior performances of learned ADCs that exceed the limits of conventional ADCs. Next, a tailored strategy is proposed to generalize the neural approximation method for arbitrary AMS circuits which are applied to the system-level designs. As a case study, the method is leveraged to transform the design of AMS peripheral circuits in in-memory computing (IMC) architectures that lead to much-improved computing performance and energy efficiency. Finally, reinforcement learning (RL) techniques are exploited to automate the design of AMS and radio-frequency (RF) circuits. Unlike prior arts, our RL agent is trained with sufficient domain knowledge of AMS circuit design, thereby beating all existing methods and experienced human designers. Particularly, our RL method is applied to solve the two most challenging problems of AMS circuits at the schematic level: 1) finding device parameters to meet given circuit specifications, and 2) performing design space exploration to achieve the best figure-of-merit (FoM) of a given circuit.

Our exploration of the automated design of AMS and RF ICs demonstrates the great potential of ML techniques to bring us closer to a future where circuit designers can be assisted by various learning methods to achieve faster and more efficient design of high-performance IC products.

Language

English (en)

Chair

Xuan X. Zhang

Committee Members

Ayan A. Chakrabarti, Shantanu S. Chakrabartty, Chuan C. Wang, Roger R. Chamberlain,

Available for download on Thursday, August 17, 2023

Share

COinS