Intent Interpretation at RIC Timescales: Jev Decision Models versus Large Language Models in 6G Open RAN
Jev-1.13.0 meets a one-second Open RAN intent budget far more often than hosted LLMs.
The paper compares Jev-1.13.0 and two other decision models with hosted large language models on RANIntent v1, in closed-loop ns-3 simulation and on a real A1 and E2 path. Median interpretation takes 0.286 to 2.35 seconds, against under 25 ms for A1 and E2 transfer. Jev-1.13.0 meets the 1-second near-real-time budget on 99.8% of calls, while two hosted LLMs meet it on 17.9% and 0%. Ideal enforcement shifts the affected-class SLA violation by 3.96 percentage points, and no hosted LLM showed a resolved gain over Jev-1.13.0 at the reported base point.
- Jev-1.13.0 meets the 1-second budget on 99.8% of calls.
- Two hosted LLMs meet that budget on 17.9% and 0%.
- Median interpretation latency is 0.286 to 2.35 seconds.
- A1 and E2 transfer stays under 25 milliseconds.
- Two interpreters saturate their queues at 2 intents per second.
- Products
- RANIntent v1
- AI models
- Jev-1.13.0
From the original article
1 min read at huggingface.co
Intent-based Open RAN needs an interpreter that turns intents into A1 policies within the loop of the RAN intelligent controller (RIC). Decision models such as Jev-1. 13.
By Delong Li, Xu Wang, Haochen Gong · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.