Algorithms for post-silicon validation and debug of radio-frequency, analog, and mixed-signal circuits and systems

Abstract

This thesis makes inroads toward a unified paradigm for the verification of hierarchical dynamical models for analog, mixed-signal, and radio-frequency circuits and systems and provides various numerical and statistical algorithms for the abstraction of AMS circuits, for quantifying dynamics “lost” in the process of abstracting, for the provision of test vectors for composite AMS systems, and for enabling a unified fault localization scheme. The primary motivation for the work is demand for techniques which leverage available system models without requiring either absolute accuracy or completeness in order to generate improved system-level tests. Methodologies are provided which benefit pass/fail production testing as well as pre- and post-silicon debugging and diagnostic testing of AMS systems. The thesis makes its contribution on three main trajectories, each informed primarily by optimization, reiforcement learning, and the design-of-experiements, in turn. Various abstract dynamical models including Volterra Filters and electronic systems including phase-locked-loops, data converters, programmable amplifiers, and radio-frequency transceivers serve as subjects for experimentation.