Quantum Reinforcement Learning Assisted Spectrum Allocation Framework for Sixth Generation Networks

Abstract

The Sixth Generation (6G) wireless networks are projected to enable ultra-dense connectivity, very high rates of data transmission, and dynamic use of spectrum and need smart and dynamic spectrum assignment. The classic spectrum allocation and classical reinforcement learning algorithms are not as effective as the complexity, scalability, and instant adaptability of 6 G space. To overcome these issues, this paper suggests a Quantum Reinforcement Learning Assisted Spectrum Allocation Framework that will combine quantum computing concepts and reinforcement learning to manage spectrum efficiently. The suggested model represents the network state in quantized form and then optimizes the spectrum allocation decision in real time with the help of a variational quantum circuit. The framework constantly communicates with the network environment and advances allegation policies on the foundation of reward feedback. It is shown in results of experiments that the proposed model has a spectrum allocation accuracy of 96.35%, spectral efficiency of $9.92 \mathrm{bps} / \mathrm{Hz}$ and throughput of $\mathbf{3 1 9 . 8 ~ M b p s}$ when the network is loaded to capacity. Also, the suggested method minimizes the latency to 21.6 ms and reduces the interference to a significant level in comparison with traditional methods. These findings verify that the developed quantum reinforcement learning model offers smart, adaptive, and efficient spectrum allocation that can be used in future 6G wireless communication systems.