Towards Intelligent 6G Networks: A Comprehensive Review of AI-Driven Control and Optimization

Deep Reinforcement Learning for 6G AI-RAN: A Comprehensive Survey

This article presents the first dedicated and comprehensive survey of DRL for Open AI-RAN, and reviews the foundations of model-free, model-based, offline, safe, multi-agent, federated, and transfer learning, and provides an O-RAN-aware framework for formulating RAN control problems through states, observations, actions, rewards, constraints, and temporal structure.

6G communications: empowering the future with advanced AI and ML techniques

Artificial Intelligence-Driven Assessment and Performance Evaluation of Fifth-Generation (5G) Wireless Communication Networks

The rapid evolution of Fifth Generation (5G) wireless communication has introduced new challenges in monitoring and optimising network performance due to increasing traffic demand, heterogeneous applications, and complex network architectures. This study assessed the role of artificial intelligence (AI) in 5G network performance using a review research approach. The review focused on eight major performance assessment areas, namely intelligent (Key Performance Indicator) KPI monitoring, predictive performance forecasting, QoE-centric (Quality of Experience) assessment, Automated Root Cause Analysis (ARCA), Radio Access Network (RAN) Optimization and self-tuning, network slicing performance assurance, energy efficiency assessment, and Digital Twin simulation for what-if testing. The work made use of recent peer-reviewed studies, gotten from reputable databases, including IEEE Xplore, SpringerLink, MDPI, Elsevier ScienceDirect, ACM Digital Library, and other credible publishers. From the reviewed studies, it was observed that AI techniques such as machine learning, deep learning, random forest, XGBoost, convolutional neural networks, long short-term memory networks, reinforcement learning, and digital twin technology can improve 5G network monitoring, intelligent decision-making, predictive maintenance, resource allocation, traffic management, energy efficiency, and network optimisation. The findings further revealed that AI enhances service reliability, reduces operational costs, minimizes latency, improves throughput, and supports proactive network management while ensuring compliance with Quality-of-Service requirements. From the results, it was concluded that AI is an essential technology for the performance assessment and optimization of modern 5G networks. This study will benefit telecommunication service providers, network engineers, researchers, government regulators, equipment manufacturers, and end users by demonstrating how AI can be applied to enhance the performance, quality of service, and sustainability of 5G communication networks.

Evolution of Power Allocation Techniques in NOMA: Advancing 5G Toward 6G Networks

The transition of 5G and beyond wireless networks toward intelligence-driven and autonomous operation has revitalized strong interest in Non-Orthogonal Multiple Access (NOMA) as an efficient multiple access framework. Power allocation critically governs NOMA performance, directly impacting throughput, user fairness, and SIC effectiveness. This survey presents a focused review of power allocation strategies in NOMA, with emphasis on the progression from static and optimization-based dynamic schemes to data-driven Artificial Intelligence (AI) and Machine Learning (ML) driven approaches. In contrast to conventional strategies that require instantaneous channel state information and iterative optimization, AI/ML techniques enable adaptive, scalable, and low-latency decision-making in highly dynamic and nonconvex environments. Recent advances in reinforcement learning and deep learning for NOMA power control are discussed, highlighting key challenges like imperfect CSI, inter-cluster interference, and distributed learning constraints. This survey provides a concise AI-centric analysis and identifies promising directions for a practical learning-driven NOMA power allocation framework for future wireless networks. A consolidated, critically comparative analysis of NOMA power allocation that bridges the gap between 5G practice and 6G imperatives is also presented in this survey.

AI-Enabled Autonomous Network Slicing Optimization for 6G Communication Systems

This paper presents a comprehensive framework for artificial intelligence (AI)-enabled autonomous network slicing optimization in 6G systems and investigates the application of advanced machine learning paradigms specifically deep reinforcement learning, federated learning, and generative AI to orchestrate dynamic resource provisioning, cross-slice isolation, and proactive SLA (Service Level Agreement) enforcement.

A Digital Twin-Based Deep Reinforcement Learning Framework for Adaptive Scheduling in 5G/6G Networks

The transition toward AI-native 6G networks requires intelligent, adaptive, and reliable control mechanisms capable of handling highly dynamic and heterogeneous environments. In this context, reinforcement learning has emerged as a promising approach for optimizing network performance. However, existing works often focus on algorithmic design while overlooking critical aspects such as experimental rigor, reward formulation, and reproducibility, which can significantly impact the validity of the results. This paper proposes a Digital Twin-based Deep Reinforcement Learning framework for adaptive scheduling in 5G networks, where the twin is implemented as a simulation-driven proxy. Within this framework, a Deep Q-Network agent dynamically selects transmission interval configurations to jointly optimize key Quality of Service metrics. A key contribution of this work lies in the systematic identification and correction of critical experimental issues, including reward degeneracy and hidden coupling between control variables, which are often neglected in RL-based networking studies. To ensure robustness, the proposed approach incorporates multi-seed statistical evaluation, providing reproducible and reliable performance assessment. Experimental results demonstrate that the proposed DRL agent consistently converges toward the optimal scheduling configuration and achieves stable performance across multiple runs, outperforming a tabular Q-learning baseline.