Knowledge-Intensive, Causal Reasoning for Analog Circuit Topology Synthesis in Emergent and Innovative Applications
Fanshu Jiaoa, Sergio Montanob and Alex Dobolic
Department of Electrical and Computer Engineering, State University of New York at Stony Brook, Stony Brook, NY 11794-2350, USA.
Analog circuit topology design has been difficult to automate. Topology synthesis involves searching an openended, widely extensible, and strongly discontinuous solution space. Existing algorithms cannot generate topologies beyond a constrained set of structures, or experience difficulties in evolving performance-effective yet minimal circuits. This paper proposes a new topology synthesis method that implements a design knowledge-intensive reasoning process to create novel circuit structures with all their features justified by the problem requirements. Two synthesis experiments demonstrate the capability of the method to generate circuits beyond the capabilities of existing topology synthesis algorithms.
Full Text (PDF)