QPPG: Quantum-Preconditioned Policy Gradient for Link Adaptation in Rayleigh Fading Channels
Oluwaseyi Giwa, Muhammad Ahmed Mohsin, Folarin Jubril Adesola, Muhammad Ali Jamshed
TL;DR
The paper tackles unstable convergence in reinforcement-learning–based link adaptation for Rayleigh fading channels by introducing QPPG, a quantum-preconditioned policy-gradient method that uses Fisher-information-based preconditioning to approximate the natural gradient. By formulating link adaptation as a POMDP and applying a conjugate-gradient solver with Fisher-vector products, QPPG achieves stabilized and accelerated policy updates. Empirical results show substantial gains in throughput ($+28.6\%$) and reductions in transmit power ($-43.8\%$) compared with classical baselines, across diverse channel conditions. This work demonstrates the viability of quantum-inspired, geometry-aware optimization for robust, data-efficient RL in future 6G networks, with potential extensions to multi-user scenarios and real-time implementations.
Abstract
Reliable link adaptation is critical for efficient wireless communications in dynamic fading environments. However, reinforcement learning (RL) solutions often suffer from unstable convergence due to poorly conditioned policy gradients, hindering their practical application. We propose the quantum-preconditioned policy gradient (QPPG) algorithm, which leverages Fisher-information-based preconditioning to stabilise and accelerate policy updates. Evaluations in Rayleigh fading scenarios show that QPPG achieves faster convergence, a 28.6% increase in average throughput, and a 43.8% decrease in average transmit power compared to classical methods. This work introduces quantum-geometric conditioning to link adaptation, marking a significant advance in developing robust, quantum-inspired reinforcement learning for future 6G networks, thereby enhancing communication reliability and energy efficiency.