A Faithful Binary Circuit Model with Adversarial Noise

TL;DR

This paper extends a faithful digital circuit delay model by incorporating limited non-deterministic variations, demonstrating that faithfulness is maintained and modeling power is increased, especially in handling glitch trains.

Contribution

It introduces a generalized involution delay model with controlled non-determinism, proving its faithfulness and enhanced modeling capabilities over previous deterministic models.

Findings

Faithfulness is preserved despite added non-determinism.

The generalized model better predicts real circuit behavior.

Small delay variations do not compromise model faithfulness.

Abstract

Accurate delay models are important for static and dynamic timing analysis of digital circuits, and mandatory for formal verification. However, F\"ugger et al. [IEEE TC 2016] proved that pure and inertial delays, which are employed for dynamic timing analysis in state-of-the-art tools like ModelSim, NC-Sim and VCS, do not yield faithful digital circuit models. Involution delays, which are based on delay functions that are mathematical involutions depending on the previous-output-to-input time offset, were introduced by F\"ugger et al. [DATE'15] as a faithful alternative (that can easily be used with existing tools). Although involution delays were shown to predict real signal traces reasonably accurately, any model with a deterministic delay function is naturally limited in its modeling power. In this paper, we thus extend the involution model, by adding non-deterministic delay variations (random or even adversarial), and prove analytically that faithfulness is not impaired by this generalization. Albeit the amount of non-determinism must be considerably restricted to ensure this property, the result is surprising: the involution model differs from non-faithful models mainly in handling fast glitch trains, where small delay shifts have large effects. This originally suggested that adding even small variations should break the faithfulness of the model, which turned out not to be the case. Moreover, the results of our simulations also confirm that this generalized involution model has larger modeling power and, hence, applicability.