Latency Exploitation in Wavelet-based Multirate Circuit Simulation

TL;DR

This paper introduces an improved wavelet-based multirate circuit simulation method that exploits circuit latency to use sparser grids, significantly reducing problem size and accelerating RF circuit simulations.

Contribution

The paper presents a novel algorithm that leverages latency to adapt grid density in wavelet multirate simulation, enhancing efficiency for RF circuit analysis.

Findings

Reduced simulation time for circuits with latent parts

Sparser grids lead to smaller problem size

Faster convergence in wavelet multirate simulations

Abstract

The simulation of radio frequency (RF) circuits is one of the severest problems in Design Automation: the information signal or envelope is modulated by a carrier signal with a center frequency typically in the GHz range. Due to Nyquist's sampling theorem the time steps in conventional transient analysis are prohibitively small. A technique to overcome Nyquist's bottleneck is the multirate method which reformulates the ordinary circuit's differential algebraic equations (DAEs) as a system of partial DAEs (PDAEs). In this paper further improvements of the wavelet multirate circuit simulation technique are presented. In the new algorithm we use different grids for the approximation of the solution on different circuit parts, exploiting latency. In particular, for circuits comprising latent parts the grids can be much sparser, which results in the reduction of the overall problem size and leads to a faster simulation.}