Adaptive Task Automata: A Framework for Verifying Adaptive Embedded Systems

We present a framework for modeling and analysis of adaptive embedded systems, based on the model of timed automata with tasks. The model is extended with primitives allowing modeling of adaptivity, by testing the potential schedulability of a given task, in the context of the set of currently enqueued tasks. This makes it possible to describe adaptive embedded systems, in which decisions to admit further tasks or take other measures of adaptivity is based on available CPU resources, external, or internal events. We show that this model can be encoded in the framework of timed automata, and hence that the problem is decidable. We also validate the framework, by using the Uppaal tool.