Comparative Analysis on ML-Driven Resource Optimization in 6G Networks

Abstract

The sixth generation (6G) of wireless networks introduces unprecedented challenges in resource optimization, virtualization, quality of service (QoS), and sustainability. This paper presents a comparative analysis of four workflows: (1) resource optimization, (2) resource allocation in AirBase virtualizers, (3) QoS enhancement in cloud storage using digital twin computing, and (4) constraint-based channel allocation in green cloud networks. Each workflow is examined against state-of-the-art literature, highlighting machine learning (ML), reinforcement learning (RL), and optimization techniques. Results demonstrate that hybrid ML-RL approaches outperform traditional methods in throughput, latency, and energy efficiency, while digital twin computing emerges as a novel paradigm for predictive QoS management. The study concludes with future research directions for sustainable and intelligent 6G systems. The paper provides the mobile resource optimization, network slicing in AirBase virtualizers, QoS enhancement in cloud storage via digital twins, and constraint-based channel allocation in green cloud networks. The review highlights methodologies such as hybrid deep learning (CNN-LSTM), AI-native slicing, federated digital twin deployment, and CVXPY-based optimization. Results indicate that hybrid ML-RL approaches outperform classical methods, while digital twin computing emerges as a novel paradigm for predictive QoS management. Research gaps include scalability, dataset standardization, synchronization overhead, and integration with renewable energy sources. Future directions emphasize federated learning, lightweight twin synchronization, and sustainable energy-aware optimization for intelligent 6G systems.