An Evaluation of Mixed Criticality Metric for Mechatronics Systems

In the thesis, we studied mechatronics systems which integrate tasks (applications) with different level of criticality on a common embedded system. The integration aims to reduce hardware cost (less processors and other components) and the weight and volume of the system. The power consumption of the system would be also reduced. The integration gives a lot of advantages but also creates new challenges. The main challenge of the system development is how to separate (isolate/protect) tasks with different criticality levels. Separation can be classified in two types, temporal separation and spatial separation. Temporal separation ensures that high criticality tasks can access resources with higher priority than low criticality tasks. Spatial separation is about preventing fault propagation from low criticality tasks to high criticality tasks. Many techniques that can make separation of tasks are studied in the thesis, and can be grouped into four categories: scheduling, power management, memory protection and communication protection. To select proper techniques that can improve the system the most, fault tree analysis and mixed criticality metric are employed. Fault tree analysis helps us to find courses of hazards that the system has to deal with. Then, we identify some techniques that can solve the problem of tasks separation. Mixed criticality metric is employed to evaluate these techniques. The evaluation result will help developers to select techniques. A self-balancing robot, which is simulated by SimScape (SimMechanics and SimElectronics) and TrueTime toolbox, was developed for experiments. Such techniques as scheduling, power management and communication protection were examined on the platform. Pros and cons ofthese techniques were evaluated. Finally, a number of recommendations for engineers with regards to the techniques for mixed criticality systems, based on our research, were provided. (Source code are shared in Matlab community)