Distributed Radiance Field Training for 6G-Enabled Metaverse: Requirements and Challenges

Immersive metaverse applications in future 6G networks will require efficient three-dimensional (3D) scene reconstruction. Techniques such as Neural Radiance Fields and 3D Gaussian Splatting offer high-quality representations but rely on centralized training with significant computational and communication demands. Distributed and federated approaches have been proposed to address these limitations by enabling decentralized processing across edge networks. This paper examines recent research on distributed radiance field training, highlighting open challenges such as communication overhead, client heterogeneity, and model synchronization. We outline possible directions for algorithmic improvement and propose architectural considerations aligned with 6G capabilities, including edge--cloud coordination and adaptive resource management. The aim is to clarify current limitations and contribute to a better understanding of the requirements for scalable and reliable scene understanding.