Behavioral macromodeling of high-speed drivers via compressed tensor representations

This paper addresses the behavioral modeling of digital drivers for Signal and Power Integrity co-simulations. State-of-the-art two-piece model representations are combined with a compact description of the device static characteristics. The latter are considered as multivariate mappings that are functions of the device electrical variables, and of additional parameters defining process corners and device settings. Overall model complexity is reduced through a compressed tensor representation obtained via a high-order singular value decomposition. Several application examples demonstrate the feasibility and the advantages of the proposed approach.