Using quantum annealing to generate test cases for cyber-physical systems
Hugo Araujo, Xinyi Wang, Mohammad Mousavi, Shaukat Ali
TL;DR
The paper tackles efficient test-case generation for Cyber-Physical Systems (CPS) by harnessing quantum annealing to guide mutation-based fuzzing. It encodes the test-case mutation problem as a Quadratic Unconstrained Binary Optimisation (QUBO) and solves it using a D-Wave quantum annealer, incorporating problem decomposition to fit hardware limits. Through two CPS case studies, it demonstrates that quantum annealing yields faster test-case generation with fault-detection performance comparable to state-of-the-art classical baselines such as simulated annealing and NSGA-II, while highlighting the importance of embedding and sub-problem strategies. The work provides both empirical and theoretical analyses of efficiency and discusses threats to validity, suggesting quantum annealing as a promising direction for software testing in safety-critical domains as hardware evolves. Overall, the results indicate that QA can enhance CPS test-generation workflows, offering practical gains in speed with competitive effectiveness.
Abstract
Quantum computing has emerged as a powerful tool to efficiently solve computational challenges, particularly in simulation and optimisation. However, hardware limitations prevent quantum computers from achieving the full theoretical potential. Among the quantum algorithms, quantum annealing is a prime candidate to solve optimisation problems. This makes it a natural candidate for search-based software testing in the Cyber-Physical Systems (CPS) domain, which demands effective test cases due to their safety-critical nature. This work explores the use of quantum annealing to enhance test case generation for CPS through a mutation-based approach. We encode test case mutation as a binary optimisation problem, and use quantum annealing to identify and target critical regions of the test cases for improvement. Our approach mechanises this process into an algorithm that uses D-Wave's quantum annealer to find the solution. As a main contribution, we offer insights into how quantum annealing can advance software testing methodologies by empirically evaluating the correlation between problem size, hardware limitations, and the effectiveness of the results. Moreover, we compare the proposed method against state-of-the-art classical optimisation algorithms, targeting efficiency (time to generate test cases) and effectiveness (fault detection rates). Results indicate that quantum annealing enables faster test case generation while achieving comparable fault detection performance to state-of-the-art alternatives.