3D Gaussian-based Immersive Media Streaming in Networked Extended Reality

3D Gaussian Splatting (3DGS) has emerged as a revolutionary representation for immersive media, offering unprecedented visual quality and efficient rendering. However, streaming 3DGS content presents unique challenges due to its complex parameter space and tight coupling between primitives. This doctoral research aims to address these challenges through three main themes: (i) developing scalable 3DGS representations that enable progressive transmission and flexible rendering, (ii) creating comprehensive quality assessment frameworks that consider both spatial-temporal coherence and multi-attribute distortions, and (iii) designing robust error control strategies for real-time streaming applications. Our research ambitions to advance the state-of-the-art in immersive media representation and delivery, with potential applications in virtual reality, telepresence, and digital twin systems.