Experimental Study of Low-Latency Video Streaming in an ORAN Setup With Generative AI
Andreas Casparsen, Van-Phuc Bui, Shashi Raj Pandey, Jimmy Jessen Nielsen, Petar Popovski · IEEE Networking Letters · 2026
Current feedback-based congestion control methods, such as probe-and-adapt solution, for live video streaming react after it occurs, causing buffering and latency spikes. We introduce a proactive semantic control channel enabling coordination between Open Radio Access Network (ORAN) xApp, Mobile Edge computing (MEC), and User Equipment (UE) for seamless mobile video streaming. When the transmitting UE experiences poor Uplink (UL) conditions, the MEC proactively instructs to downscale video based on low-level RAN metrics–such as millisecond SNR updates, preventing buffering before it fully manifests. A Generative AI (GAI) module at the MEC reconstructs high-quality frames from downscaled video before forwarding them over the typically stronger Downlink (DL). Experiments on a live ORAN testbed with 50 video streams show reduced latency tails and up to 4 dB PSNR and 15 Video Multi-Method Assessment Fusion (VMAF) gains over reactive congestion control. The proactive control eliminates latency spikes over 600 ms, demonstrating effective cross-layer coordination for latency-critical streaming.