Composing Functional and State-Based Performance Models for Analyzing Heterogeneous Real-Time Systems

Abstract

We show that by appropriately composing these two classes of models it is possible to leverage on their respective advantages.To this end, we propose an interface between components that are modeled using Real-Time Calculus [Chakraborty, Künzli and Thiele, DATE 2003] and those that are modeled using Event Count Automata [Chakraborty, Phan and Thiagarajan, RTSS 2005]. The resulting modeling technique is as expressive as Event Count Automata, but is amenable to more ef cient analysis. We illustrate these advantages using a number of examples and a detailed case study.

Proceedings of the 4th Analytic Virtual Integration of Cyber-Physical Systems Workshop December 3 Vancouver Canada

“…Furthermore, some systems may contain a mixture of state-based and stateless components; to efficiently support this case, it would be useful to have a way to interconnect components of both types. It seems interesting to explore hybrid techniques, along the lines of [33] and [24].…”

Causality closure for a new class of curves in real-time calculus

“…The causality problem has received surprisingly little attention in the Real-Time Calculus community, although many of the existing approaches which connect Real-Time Calculus to other formalisms (e.g. [9,8,13]) did produce, or have problems with non-causal pairs of curves.…”

“…This is the case when using techniques like abstract interpretation, model-checking algorithms with a timeout like in [2], or when abstractions are performed on the model before the computation [1]. It can also happen when an algorithm computes only part of the points of an arrival curve, like done in [13].…”

“…Some works proposed RTC extensions using Event Count Automata [13], Timed Automata [9,10,15,1], and Synchronous Programs [2]; all of those approaches perform the analysis using some formal verification tool (model-checking, abstract interpretation) and they all face the causality problem [3] in a way or another. They express the input arrival curves as a non-deterministic generator, or an observer of events that encodes the constraints of the curves.…”

“…Conversely, they can also be produced by non-exact computations: e.g. when a curve is computed after abstracting the system [1], using abstract interpretation [2], or using a non-complete model-checking algorithm [13]. It is therefore useful in some cases, mandatory in others, to solve this issue by transforming non-causal curves into causal ones.…”