Demo: A Reinforcement Learning-based Flexible Duplex System for B5G with Sub-6 GHz

Abstract

In this paper, we propose a reinforcement learning-based flexible duplex system for B5G with Sub-6 GHz. This system combines full-duplex radios and dynamic spectrum access to maximize the spectral efficiency. We verify this method's feasibility by implementing an FPGA-based real-time testbed. In addition, we compare the proposed algorithm with the result derived from the numerical analysis through system-level evaluations.