Year: 2015
Type: dissertation
Abstract: ENGLISH ABSTRACT: In order to facilitate the creation of complex and robust RSFQ digital logic \ncircuits an extensive library of electronic design automation (EDA) tools is a \nnecessity. It is the aim of this work to introduce various methods to improve \nthe current state of EDA in RSFQ circuit design. \nFirstly, Monte Carlo methods such as Latin Hypercube sampling and Sobol \nsequences are applied for their variance reduction abilities in approximating \ncircuit yield. In addition, artificial neural networks are also investigated for \ntheir applicability in modeling the parameter-yield space. \nSecondly, a novel technique for circuit functional testing using automated \nstate machine extraction is presented, which greatly simplifies the logical verification \nof a circuit. This method is also used, along with critical timing \nextraction, to automatically generate Hardware Description Language(HDL) \nmodels which can be used for high level circuit design. \nLastly, the Greedy Local search, Simulated Annealing and Genetic Algorithm \nmeta-heuristics were statistically compared in a novel manner using a \nyield model provided by artificial neural networks. This is done to ascertain \ntheir performance in optimising RSFQ circuits in relation to yield. \nThe variance reduction techniques of Latin Hypercube Sampling and Sobol \nsequences were shown to be beneficial for the use with RSFQ circuits. For \noptimisation purposes the use of Simulated Annealing and Genetic Algorithms \nwere shown to improve circuit optimisation for possible multi-modal search \nspaces. An HDL model is also successfully generated from a complex RSFQ \ncircuit for use in high level circuit design which includes critical timing and \npropagation latency. \nAll the techniques presented in this study form part of a software library \nthat can be further refined and extended in future work.
Author: Louis Christiaan Muller
Language: English
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W1910332527