Simulation-based Design of Critical Systems via Black-Box Optimization and Statistical Model Checking

October 7, 2024

Marco Esposito (LACL/Verimag)

Designing provably robust cyber-physical systems is an ambitious challenge, which usually requires employing multiple techniques from different branches of engineering and mathematics.
When the system under design is particularly complex, and there is uncertainty in the system or the environment, classical white-box methods from systems engineering are not feasible, and simulation-based approaches are the only viable option.
In this talk, I will present my work in optimization-based methods for system design that exploit Statistical Model Checking to provide formal guarantees on the robustness of the solution.
I will show different flavors of this approach in three application fields: the design of networks of anti-drone sensors in critical areas, the synthesis of robust controllers for complex industrial CPSs, and the model-based optimization of personalized therapies for cancer.