Recent Advances in Deep Learning with Optimization-Based Task Scheduling and Computing Resource Allocation for VR Video Services in Advanced 6G Networks: A Systematic Review

Abstract

The emergence of 6G networks is set to transform immersive applications such as Virtual Reality (VR) services, which require ultra-low latency, high bandwidth, and efficient resource utilization. However, VR applications generate massive data streams and demand real-time processing, making task scheduling and resource allocation significant challenges. Traditional optimization techniques often fail to handle the dynamic and stochastic nature of 6G environments, prompting the adoption of deep learning-based solutions. This review examines recent advances in frameworks that integrate deep learning and optimization for joint task scheduling and resource allocation in VR-enabled 6G networks. Deep Reinforcement Learning (DRL) has proven effective for dynamic decision-making, enabling intelligent task offloading and adaptive resource management in edge computing scenarios. Additionally, Lyapunov optimization techniques provide stability guarantees and efficient latency control by transforming complex problems into queue-based models. Hybrid approaches combining DRL and Lyapunov optimization have demonstrated improved performance in terms of latency reduction, adaptability, and Quality of Experience (QoE). This review highlights recent developments, compares methodologies, and identifies challenges such as scalability, computational overhead, and energy efficiency in next-generation 6G VR systems.