ML-OLLA: Link Adaptation for Real-Time Traffic in 5G/6G NTN Systems
Reliable low-latency connectivity for real-time applications is a key requirement for 5G and future 6G non-terrestrial networks. Low Earth Orbit satellites enable global coverage but impose long propagation delays, rapid channel variations, and delayed feedback, which challenge conventional link adaptation algorithms designed for terrestrial networks. To address this problem, the paper proposes a Multi-Level Outer-Loop Link Adaptation (ML-OLLA) algorithm that assigns a dedicated target block error rate and an independent adaptation process for each transmission attempt. With ML-OLLA, early transmission attempts use higher-rate modulation and coding schemes (MCSs), while later ones use more robust MCSs. The core contribution of this work is a low-complexity data-driven parameter selection method that determines ML-OLLA parameters. Numerical results obtained with ns-3 show that the proposed ML-OLLA algorithm increases the system capacity by up to four times and reduces the minimum supported latency budget. In particular, for the 99.9% reliability target, ML-OLLA reduces the minimum supported latency budget from 60 ms to 40 ms compared with the baseline OLLA.