Search based Model in the Loop Testing for Cyber Physical Systems

Increasing the safety of software controlled complex systems, such as automotive vehicles that use multiple, distributed electronic control units to control the safe operation of systems such as engines, brakes and airbags, requires extensive testing, which can be time consuming and expensive. We can optimize the test process with respect to testing depth and incurred costs by applying automated testing in conjunction with hardware-in-the-loop, software-in-the-loop, and model-in-the-loop simulation. In this paper, we propose a search-based approach using a Discrete Genetic Algorithm to generate test cases for continuous controllers at the MiL level, discretizing the input variables domains. We identify a set of requirements for the desired behavior and search for test cases that break the requirements. We illustrated our approach on a cruise control system for a hybrid propulsion bicycle (e-Bike) case study and performed several experiments: to find the best configuration for the discrete genetic algorithm and to compare the results with the continuous genetic algorithm. The results derived from the experiments show that the proposed approach is effective in model-in-the-loop testing for cyber-physical systems.