Edge Computing Empowered Holographic Video Communication: A Multi-Objective Hierarchical Reinforcement Learning Approach

As a novel future communication paradigm, holographic video communication aims to create an unprecedented immersive interactive experience by recreating high-definition, fine-grained images of real-world objects. Compared to traditional real-time communication services, holographic video communication imposes unprecedentedly high network bandwidth and computing power standards. Although edge computing technology provides possibilities for alleviating these demands, its internal dynamics pose a significant barrier to the performance of holographic video communication. To address this, we introduce an innovative multi-objective hierarchical reinforcement learning (MOHRL) framework designed to optimize the adaptability of holographic video communication in dynamic edge network environments. Our innovative approach involves designing a multimodal spatial-temporal transformer model that can sensitively capture subtle changes in communication tasks and network conditions, providing robust information support for the hierarchical reinforcement learning framework. The high-level policy focuses on learning and setting stage-based goals, guiding the low-level policy to make real-time decisions in complex network environments and ensuring holographic video communication services' smooth and stable operation. Furthermore, we establish a rigorous comparative experiment scheme to systematically validate the performance gains of our proposed method in practical applications, highlighting its superiority through experimental results. Finally, this article explores future research trends and challenges in holo-graphic video communication, laying a theoretical and practical foundation for future researchers.