The AI-Native RAN: What's Standardized, What's Deployed
Mapping 3GPP's AI for the RAN from Rel-18 to Rel-20 — and separating standardized reality from AI-RAN hype.
Abstract
3GPP's Rel-18 study mapped three air-interface AI use cases; Rel-19 made beam prediction and positioning normative; Rel-20 took two-sided CSI compression into normative work. The paper separates 3GPP's AI-for-RAN from the AI-RAN Alliance's GPU-convergence agenda and reads the 6G air interface soberly.
Headline finding
3GPP standardized just three air-interface AI use cases — only beam prediction and positioning are normative (Rel-19)
Get the full paper
7 pages, 1.4 MB. Add your email to unlock the PDF — plus the next whitepaper the day it publishes. Unsubscribe in one click.
Start with the article
This paper deepens one of our guides. If you want the shorter version first, read that.
Read the guide