No abstract

An interlocking system monitors the status of the objects in a railway yard, allowing or denying the movement of trains, in accordance with safety rules. The high number of complex interlocking rules that guarantee the safe movements of independent trains in a large station makes the verification of such systems a complex task, which needs to be addressed in conformance with EN50128 safety guidelines.In this paper we show how the problem has been addressed by a manufacturer at the final validation stage of production interlocking systems, by means of a model extraction procedure that creates a model of the internal behaviour, to be exercised with the planned test suites, in order to reduce the high costs of direct validation of the target system. The same extracted model is then subject to formal verification experiments, employing an iterative verification process implementing slicing and CEGAR-like techniques, defined to address the typical complexity of this application domain.

Abstract. This paper reports the story of the introduction of formal methods in the development process of a railway signaling manufacturer. The first difficulty for a company is due to the many different formal methods proposals around; we show how this difficulty has been addressed and how the choice of a reference formal specification notation and of the related tools has been driven by many external factors related to the specific application domain, to the company policies, to european regulations. Cooperation with University has been fundamental in this process, which is now at the stage in which internal acceptance of the chosen formalisms and tools is established.

No abstract

Introduction of formal model-based practices into the development process of a product in a company implicates changes in the verification and validation activities. A testing process that focuses only on code is not comprehensive in a framework where the building blocks of development are models, and industry is currently heading toward more effective strategies to cope with this new reality. This paper reports the experience of a railway signalling manufacturer in changing its unit level verification process from code-based testing to a two-step approach comprising model-based testing and abstract interpretation. Empirical results on different projects, on which the overall development process was progressively tuned, show that the change paid back in terms of verification cost reduction (about 70%), bug detection, and correction capability.