Key points are not available for this paper at this time.
For a robust circuit design, statistical variations of device parameters must be considered. A Monte Carlo (MC) simulation is the standard approach to get a detailed picture for the statistical variation of a circuit's performance. However, there is a high number of iterations necessary for getting reliable statistical data. The Noise-Based Variability Approach (NOVA) has been proposed as alternative and a numerically very efficient method. It is based on a noise analysis in a circuit simulation with fluctuating parameters of a compact model and allows estimating the statistical variability of state variables in a circuit from only one simulation run. In this paper, NOVA is reviewed in comparison to MC simulations for the case of circuits with organic thin-film transistors including parameter fluctuations and measurements on single devices. Furthermore, the potential of NOVA for the simulation of the inference process in memristive crossbar arrays of neuromorphic computing structures is discussed.
Kloes et al. (Wed,) studied this question.