Long-Term Energy-Efficient Duplex Mode Optimization for Cell-Free Massive MIMO Systems With Network-Assisted Full-Duplex

The network-assisted full-duplex (NAFD) system enables simultaneous uplink (UL) and downlink (DL) transmission through dynamic allocation of UL and DL remote antenna units (RAUs), thereby greatly improves the system’s performance. This article explores the optimization of long-term energy efficiency (EE) for the NAFD cell-free massive multiple-input multiple-output (CF-mMIMO) system. In order to maximize long-term EE while maintaining network stability, an Energy-Efficient Lyapunov-guided duplex mode optimization scheme (EE-LyDMO) is proposed subject to data queue stability, average power consumption limit, and instantaneous quality of service (QoS) requirements. By introducing the Lyapunov framework, the original stochastic problem is converted into an equivalent instantaneous deterministic problem. Furthermore, a reinforcement learning (RL)-based duplex mode optimization algorithm is introduced to address the instantaneous problem. Theoretical analysis and simulation results showcase that the proposed EE-LyDMO can attain a substantial improvement in EE performance while upholding network stability, and the EE-backlog tradeoff is derived as <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$[\mathcal {O}(1 / V), \mathcal {O}(V)]$ </tex-math></inline-formula> with V as a control parameter.