QoR Model Translation and Evaluation
Bachelor Thesis, Seminar Thesis, Master Thesis
In mission-critical applications such as autonomous driving, the speed of response of a distributed system matters. Most of these applications do not require a 100% accurate response, but are limited by time constraints. Therefore, Quality of Result (QoR) adaptation is a suitable way to meet timing constraints.
Current models have problems with reconfiguring possible application paths, plus current approaches may not be the most efficient.
Goals
- Automated conversion between different models
- Dealing with difficult situations such as:
- Parallel execution paths
- Different types of sub-tasks
- Reconfiguration of the model
Vision
The QoR is a well-working tool to find a suitable result in expected time. To improve the execution time and the time needed to compare different paths, we can automatically translate an application structure between the different models. We can also determine whether the graph or the feature model is more efficient for decision-making and dealing with reconfiguration.
