A Link Adaptive Intelligent Modulation and Coding Strategy Based on Deep Reinforcement Learning

To address the challenge of coordinated optimization for reliability and efficiency in 5 G adaptive modulation and coding, this paper proposes an intelligent decision-making strategy based on the Dueling Deep Q-Network. By constructing a dual-branch network architecture that decouples state value and action advantage, dynamic perception of time-varying channel states and adaptive optimization of modulation coding schemes are achieved. In scenarios with signal-to-noise ratio fluctuations ranging from −5 dB to 35 dB, the proposed strategy significantly reduces the signal-to-noise ratio requirement corresponding to the bit error rate threshold and improves spectral efficiency compared with traditional channel quality indicator-modulation and coding scheme (CQI-MCS) lookup methods and basic Deep Q-Network algorithms. This study validates the effectiveness of deep reinforcement learning in breaking through fixed threshold limitations and realizing dynamic adaptive modulation and coding optimization, providing new insights for the coordinated adaptation of 5 G URLLC and eMBB services.