Decreasing maintenance costs by introducing formal analysis of real-time behavior in industrial settings

• Authors: Anders Wall;Johan Andersson;Christer Norström
• Venue: ISoLA'04 Proceedings of the First international conference on Leveraging Applications of Formal Methods
• Year: 2004

A common problem with long-lived large industrial software systems such as telecom and industrial automation systems is the increasing complexity and the lack of formal models enabling efficient analyses of critical properties. New features are added or changed during the system life cycle and it becomes harder and harder to predict the impact of maintenance operations such as adding new features or fixing bugs. We present a framework for introducing analyzability in a late phase of the system's life cycle. The framework is based on the general idea of introducing a probabilistic formal model that is analyzable with respect to the system properties in focus, timing and usage of logical resources. The analyses are based on simulations. Traditional analysis method falls short due to a too limited modelling language or problems to scale up to real industrial systems. This method can be used for predicting the impact caused by e.g. adding a new feature or other changes to the system. This enables the system developers to identify potential problems with their design at an early stage and thus decreasing the maintenance cost. The framework primarily targets large industrial real-time systems, but it is applicable on a wide range of software system where complexity is an issue. This paper presents the general ideas of the framework, how to construct, validate, and use this type of models, and how the industry can benefit from this. The paper also present a set of tools developed to support the framework and our experiences from deploying parts of the framework at a company.