Verifying the adaptation behavior of embedded systems

Abstract

Many complex embedded systems dynamically adapt their components, services, algorithms, and parameters to the environment. This leads to new classes of design errors, since adaptation has become an increasingly complex part of the systems' behavior. In particular, as adaptations often continuously trigger further adaptations in other components, inconsistent and unstable configurations may be reached. Formal verification, which is routinely applied in safety-critical applications, must therefore consider not only temporal and functional properties of a system, but also its ability to dynamically adapt itself according to external and internal stimuli.In this paper, we describe how the adaptation behavior of embedded systems can be modeled, specified, and verified at design time. The systems are thereby given at a high level of abstraction, where adaptation is triggered by the quality of data values. This allows to extract the relevant information in a form that can be directly used for verification. Moreover, we demonstrate how state-of-the-art model checkers can be used to formally reason about the resulting system description.