Model-driven development of critical perception components using Simulink

Modern sensor-actuator applications combine a large number of components (sensing devices, processing nodes, networks) and implementing complex interactions between them. Due to failures and other sensor inherent insufficiencies the intended control function is adversely affected. This cannot be tolerated in safety critical applications. In some cases replication and voting may be possible. But this is no general solution. Many sensors cannot be replicated because of cost or their operating principles. This demands other failure detection and handling mechanisms to meet application requirements. Model-driven development techniques can be exploited here to adjust the failure handling to the needs of the application. In this paper, we propose a Simulink framework that supports the entire development chain. This includes a new description technique, design verification using regular expressions in combination with a model generator. In contrast to existing approaches, our scheme applies one modeling concept and one development environment throughout the entire process.