A Compositional Interchange Format for Hybrid Systems: Design and Implementation

Damian Nadales

Promotores: prof.dr.ir. J.E. Rooda (TU/e) and prof.dr. J.C.M. Baeten (TU/e, CWI)
Co-promotor: dr.ir. D.A. van Beek (TU/e)
Technische Universiteit Eindhoven
Date: 29 November, 2012

Summary

The design of large industrial controlled systems is a difficult task, which calls for a model-based design approach. For this, different formalisms exist. Each of these formalisms addresses a specific set of problems, and has its own set of features. Moreover, several formalisms and tools are used in each stage of the system development.

As a consequence, there is a conspicuous need for integrated tool support for the design of large complex controlled systems, from the first concept to the implementation, and further on, over their entire life cycle.

The Compositional Interchange Format (CIF) has been developed during this research to serve as an interchange format between different formalisms. CIF is an automata-based formalism, that allows to model and simulate hybrid systems. The language incorporates process algebraic operations, and has a formal semantics, based on the Structured Operational Semantics (SOS) framework. The semantics is shown to be compositional and we have proven that it preserves important algebraic properties, which express our intuition about the behavior of the language operators.

CIF was extended with hierarchy (HCIF), to support the development of systems through stepwise refinement. The semantics of HCIF is given by means of SOS rules, and is defined in a compositional manner, by referring only to the transition system of the substructures, and not to their syntactic representation. This compositional introduction of hierarchy allows us to keep the semantics of the HCIF operators almost unchanged with respect to the CIF original semantics. A case-study on a patient support system is modeled in HCIF to show its applicability.

Based on the formal description of CIF, we developed a simulator that conforms to the language semantics. We use the SOS specification of the language to obtain a new set of so-called symbolic rules. These rules contain the predicates that are required during simulation to compute time delays, and action updates. In this way, we present a rigorous method, which given the semantic specification of a complex language, allows us to obtain a simulator for models written in that language.

We defined a linearization process for CIF terms, to allow the elimination of operators that are not natively supported in other languages, and to facilitate tool reuse. The linearization algorithm is obtained through a stepwise refinement of the original CIF SOS rules. As a result, we show how the semantic specification of the language can guide the implementation of such a procedure, yielding a simple proof of correctness.

To enable the verification of timed CIF models, we formalized and proved a semantic-preserving transformation from CIF to Uppaal.

Finally, the tools we developed were used in a complex case-study: the design of a controller for a miniature pipeless plant. In this way we show the applicability, as a proof of concept, of the toolchain developed in the FP7-MULTIFORM project. This toolchain involves controller synthesis, verification, and hybrid-systems simulation.

In this thesis we show how formal semantics can be used not only for specifying mathematically a language, but also for developing tools and model transformations for it. However, working with such a degree of formalization requires the development of tool support for assisting this process. Otherwise the process is not maintainable nor scalable. In hindsight it is clear now that besides the mathematical correctness, the modeling convenience of a modeling formalism is crucial for its adoption. However in this thesis the effort was put mostly in the former aspect.