TRUSTWORTHY STRUCTURAL SYNTHESIS AND EXPERT KNOWLEDGE EXTRACTION WITH APPLICATION TO ANALOG CIRCUIT DESIGN

摘要

A system and method that does trustworthy multi-objective structural synthesisof analogcircuits, and extracts expert analog circuit knowledge from the resultingtradeoffs. Thesystem defines a space of thousands of possible topologies via ahierarchically organizedcombination of designer-trusted analog building blocks, the resultingtopologies areguaranteed trustworthy. The system can perform a search based on a multi-objectiveevolutionary algorithm that uses an age-layered population structure tobalance explorationvs. exploitation, with operators that make the search space a hybrid betweenvector-basedand tree-based representations. A scheme employing average ranking on Paretofronts isused to handle a high number of objectives. Good initial topology sizings arequicklygenerated via multi-gate constraint satisfaction. To explicitly capture expertanalog designknowledge, data mining is employed on the sized circuits to: automaticallygenerate adecision tree for navigating from performance specifications to topologychoice, to do globalnonlinear sensitivity analysis, and to generate analytical models ofperformance tradeoffs.