IEEE Open Journal of Antennas and Propagation
Quantum Optimization for Reconfigurable Intelligent Surface Design: Trainability and Barren Plateau Analysis
Emanuel Colella, Luca Bastianelli, Valter Mariani Primiani, Franco Moglie, Zhen Peng, Gabriele Gradoni
Abstract
Reconfigurable intelligent surfaces (RISs) have emerged as a promising technology in wireless communications, offering dynamic control over electromagnetic propagation to face challenges such as latency, interference, and multipath fading. However, optimizing RIS configurations in real time remains computationally prohibitive due to the exponential growth of the solution space with the number of elements. In this work, we explore quantum-based RIS optimization using the quantum approximate optimization algorithm (QAOA) within a model-driven framework based on the Sherrington–Kirkpatrick (SK) Hamiltonian. This formulation exploits superposition, entanglement, and quantum interference to enable parallel exploration of the configuration space, while avoiding the need for large training datasets and minimizing the risk of overfitting. To validate the method, we adopt a one-to-one mapping between RIS elements and qubits, in order to test the effectiveness of QAOA combined with SK model for RIS optimization. Our results reveal that barren plateaus—regions of vanishing gradient—emerge around 12–14 qubits, marking a fundamental scalability barrier under direct encoding. This diagnostic insight provides a baseline for future improvements and underscores the need for more scalable encodings and advanced training strategies to enable efficient RIS optimization in practical, large-scale wireless environments.