A situation-aware task model for adaptive real-time systems

Real-time systems are usually well-defined and operate based on a specific task model defined during system design. However, the system can interact with different objects from its environment at runtime and needs to guarantee its operational as well as timing behavior even in adverse environmental situations. Uncertainties in the system environment impose challenges on assuring the runtime behavior during system design. The system needs to adapt to different environmental situations which require the task model to consider the execution of adaptive tasks which can be activated in response to environmental events. We present an operational environment model that characterizes environmental situations of the real-time system and identifies the adaptive tasks needed to be activated at runtime. The adaptive tasks can be included and executed using a number of existing task models which allow non-deterministic task activation patterns. We present a situation-aware task model which efficiently maps the environmental events to (reduced) adaptive tasks. To demonstrate the applicability and usability of the proposed situation-aware task model, we perform the experimental analysis using two case studies: an automotive situation-aware task model, and an unmanned aerial vehicle situation-aware task model. The experimental results of our work show that the constructed situation-aware task model contains a maximum of nine vertices and 68 edges, provides an improvement in terms of scheduling overhead and in adaptation time (with respect to the considered existing task models).