6G • RAN

DeepJSCC-XR

Deep Joint Source-Channel Coding for Immersive XR

End-to-end neural autoencoders that compress and transmit volumetric 3D point clouds directly over fading wireless links without digital compression artifacts.

Technical Explanation

Transmitting raw 3D holographic video and point clouds requires gigabits per second. Traditional codecs (like MPEG-I or H.265) cause severe blocky compression artifacts or complete video freezing whenever wireless packets drop. DeepJSCC-XR trains neural networks to directly map 3D point cloud coordinates and color textures into analog wireless symbols, ensuring photorealistic rendering and graceful degradation even when signal strength drops.

Key Functions

  • End-to-end neural mapping of volumetric 3D meshes into wireless channel symbols
  • Eliminating blocky video artifacts and screen freezing during wireless channel fades
  • Sub-10 millisecond motion-to-photon end-to-end delivery latency for AR/VR headsets
  • Graceful visual degradation maintaining immersion even in extreme low-SNR zones
  • Feature preservation prioritizing human facial expressions and hands during conversations
Specifications
IEEE Transactions on Multimedia, 3GPP Rel-20 Study on Immersive Media
Interfaces
JSCC-XR-StreamVolumetric-Air

Related 6G Concepts

Want to memorize 6G concepts like this one?
Study it with SuperMemo SM-2 spaced repetition flashcards.
Practice 6G Flashcards