Mobile communication technology has evolved significantly from 1G analog systems to 5G intelligent networks, focusing on improvements in spectral efficiency, network capacity, and adaptability. Currently, 5G encounters challenges such as limited spectrum resources and increased bandwidth pressure, especially in high-density user environments. While traditional Non-Orthogonal Multiple Access (NOMA) technologies, like SCMA, and Device-to-Device (D2D) communication enhance spectrum reuse, they still face issues with dynamic resource allocation and cross-layer collaboration. This paper introduces an intelligent resource allocation framework leveraging Deep Reinforcement Learning (DRL) for dynamic power control and interference coordination through multi-agent collaboration. A Markov decision process model is developed, and a distributed DRL algorithm is created to optimize local and global performance in cellular networks. Experiments show that DRL-driven SCMA codebook scheduling can improve spectral efficiency by 20% while enabling distributed interference management and network slicing optimization in D2D scenarios. Nonetheless, challenges remain in practical DRL deployment, such as online training costs and policy interpretability. Future advancements will involve integrating DRL with sixth-generation (6G) technologies like intelligent reflecting surfaces (RIS) and terahertz beamforming, fostering a shift towards cognitive communication systems with autonomous perception and global optimization.
Yan Zhang (Tue,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: