Multiple Objective Functions for Falsification of Cyber-Physical Systems
Paper in proceeding, 2020

Cyber-physical systems are typically safety-critical, thus it is crucial to guarantee that they conform to given specifications, that are the properties that the system must fulfill. Optimization-based falsification is a model-based testing method to find counterexamples of the specifications. The main idea is to measure how far away a specification is from being broken, and to use an optimization procedure to guide the testing towards falsification. The efficiency of the falsification is affected by the objective function used to evaluate the test results; different objective functions are differently efficient for different types of problems. However, the efficiency of various objective functions is not easily determined beforehand. This paper evaluates the efficiency of using multiple objective functions in the falsification process. The hypothesis is that this will, in general, be more efficient, meaning that it falsifies a system in fewer iterations, than just applying a single objective function to a specific problem. Two objective functions are evaluated, Max, Additive, on a set of benchmark problems. The evaluation shows that using multiple objective functions can reduce the number of iterations necessary to falsify a property.

Multiple Objective Functions

Testing

Cyber-Physical Systems

Falsification

Author

Zahra Ramezani

Chalmers, Electrical Engineering, Systems and control, Automation

Johan Lidén Eddeland

Chalmers, Electrical Engineering, Systems and control, Automation

Koen Claessen

Chalmers, Computer Science and Engineering (Chalmers), Functional Programming

Martin Fabian

Chalmers, Electrical Engineering, Systems and control, Automation

Knut Åkesson

Chalmers, Electrical Engineering, Systems and control, Automation

IFAC-PapersOnLine

24058963 (eISSN)

Vol. 53 4 417-422

15th IFAC Workshop on Discrete Event Systems, WODES 2020
Rio, Brazil,

Roots

Basic sciences

Subject Categories

Embedded Systems

Computer Science

Computer Vision and Robotics (Autonomous Systems)

DOI

10.1016/j.ifacol.2021.04.040

More information

Latest update

6/10/2021