Gulbahar and Colleagues Introduce $2^{m}$-Phase Dictionary for RIS Power Optimization

Researchers Burhan Gulbahar, from the Yasar University, and colleagues, have developed a new method for optimising power in reconfigurable intelligent surface (RIS) systems. Their design utilises a 2^M-phase dictionary for offline QAOA angle optimisation, bounding spin-Hamiltonian interaction order and demonstrating performance approaching a near-optimal classical reference for RIS channels with up to 16 elements. The method suggests a pathway towards achieving near-optimal performance with large-scale RIS deployments on future quantum computers. The team tackles the computational challenge of discrete-phase optimisation, typically NP-hard, and the limitations of applying the quantum approximate optimisation algorithm (QAOA) to RIS, such as barren plateaus and the need for scalable benchmarks.

Near-optimal reconfigurable intelligent surface optimisation via bounded spin-Hamiltonian modelling

A 32-phase dictionary enabled near-optimal performance, reaching 99% of a classical reference point, for reconfigurable intelligent surface systems with up to 16 elements. This feat was previously unattainable due to the computational complexity of discrete-phase optimisation. Reconfigurable intelligent surfaces represent a paradigm shift in wireless communication, offering the ability to dynamically control the wireless propagation environment. Unlike traditional relaying or MIMO techniques, RIS do not perform active signal processing, instead relying on passive reflection of signals. Optimising the phase shifts of these reflecting elements is crucial to maximise signal strength and improve network performance. However, the discrete nature of phase control introduces a combinatorial optimisation problem that becomes exponentially more difficult as the number of RIS elements, denoted by M, increases. This is because the search space grows as 2^M, rendering exhaustive search impractical even for moderately sized RIS.

The advancement overcomes limitations of the quantum approximate optimisation algorithm, or QAOA, when applied to RIS, specifically the challenges of barren plateaus and the need for scalable benchmarks. QAOA is a hybrid quantum-classical algorithm designed to find approximate solutions to combinatorial optimisation problems. While promising in theory, QAOA suffers from the ‘barren plateau’ phenomenon, where the gradient of the cost function vanishes exponentially with the problem size, hindering the training process. Furthermore, evaluating the performance of QAOA requires scalable benchmarks, which are currently lacking for RIS optimisation. The researchers circumvent these issues by employing a pre-computed dictionary of angles, effectively reducing the complexity of the optimisation problem. This dictionary approach allows for offline optimisation of the QAOA parameters, mitigating the impact of barren plateaus and enabling a more efficient search for near-optimal solutions.

Bounding the spin-Hamiltonian interaction order ensured viability even as the number of elements increases, and order-2 modelling surpassed the performance ceiling of order-4 models. The spin-Hamiltonian model is a mathematical framework used to represent the interactions between the RIS elements. Higher-order interactions (e.g., order-4) capture more complex relationships but also increase the computational burden. By limiting the interaction order to 2, the researchers were able to maintain computational tractability while still achieving excellent performance. This suggests that capturing the dominant interactions between RIS elements is sufficient for near-optimal optimisation, and that incorporating higher-order interactions may not yield significant improvements. This is a crucial finding, as it simplifies the optimisation problem and reduces the computational resources required. Pre-calculated angles transferred consistently to various channel conditions, including Rayleigh, Rician, cascaded double-fading, and spatially-correlated environments, across five and twelve element systems. This robustness to channel variations is essential for practical deployment, as real-world wireless channels are often complex and time-varying. Analytical, state-vector, matrix-product-state and Pauli-path simulations were employed to validate the approach. These diverse simulation techniques provide a comprehensive assessment of the method’s accuracy and reliability. Current results assume ideal conditions and do not yet demonstrate performance in real-world scenarios with noisy hardware or imperfect component calibration, representing a key area for future work.

Quantum limitations currently restrict scaling of optimised wireless surface performance

Gulbahar and colleagues have demonstrated a promising method for optimising power in reconfigurable intelligent surfaces, technology poised to enhance wireless signal strength and network capacity. RIS offer a cost-effective and energy-efficient solution for improving wireless coverage and capacity, particularly in challenging environments such as urban canyons or indoor spaces. However, realising the full potential of RIS requires efficient optimisation algorithms that can handle the increasing complexity of large-scale deployments. Performance degrades with larger configurations, mirroring a challenge highlighted by Colella et al. regarding barren plateaus in quantum optimisation, though the observed performance ceiling with current quantum hardware does not negate the importance of this work. The limitations of current quantum hardware, such as qubit coherence times and gate fidelities, pose a significant obstacle to scaling up QAOA-based RIS optimisation. While the researchers achieved near-optimal performance for RIS with up to 16 elements, extending this to larger configurations will require advancements in fault-tolerant quantum computing.

This offers a clear strategy for future improvement as quantum technology matures and can handle more complex calculations. The development of more robust and scalable quantum algorithms, as well as improvements in quantum hardware, will be crucial for unlocking the full potential of RIS optimisation. A viable strategy for optimising these devices, which enhance wireless signal strength by precisely controlling radio waves, has been established. Employing a pre-calculated, limited set of phase adjustments, a ‘dictionary’ of options, reduced the computational burden on a quantum algorithm. This achieved performance closely matching that of conventional methods for systems containing up to sixteen elements, representing a major step towards practical application and allowing for further investigation into the limitations of scaling with larger configurations. The use of a pre-calculated dictionary effectively transforms the optimisation problem into a lookup task, significantly reducing the computational complexity. This approach is particularly well-suited for implementation on quantum computers, where the cost of performing complex calculations is high.

The research demonstrated a method for optimising reconfigurable intelligent surfaces, devices that can enhance wireless signal strength, using a pre-calculated dictionary of phase adjustments. This technique reduces the computational demands on quantum algorithms, achieving performance comparable to conventional methods for systems with up to sixteen elements. By limiting the complexity of the quantum calculations, the researchers circumvented issues related to the scalability of quantum optimisation algorithms. Future work will focus on extending this approach to larger configurations as quantum computing technology advances.