Falsification is a simulation-based testing method for Cyber-Physical Systems (CPSs) used to find input conditions that violate formal requirements. While effective, traditional falsification techniques struggle when multiple, possibly independent probabilistic requirements must be tested under nondeterministic or flaky simulations. This paper introduces Parametric Falsification (PF), a method that combines parametric model checking and many-objective optimization to efficiently detect violations across multiple probabilistic requirements. PF shifts computational complexity offline by precomputing satisfaction constraints for all requirements and then employs optimization to explore parameter spaces during simulation. Experiments on four open-source CPS benchmarks demonstrate that PF falsifies more requirements, with higher violation severity than selected baselines, while adding negligible offline cost.
Camilli et al. (Thu,) studied this question.