Open RAN approaches for 6G Non-Terrestrial Networks

This thesis evaluates applying the O-RAN paradigm to NTNs to deliver flexible, interoperable and scalable terrestrial and non-terrestrial integrated 6G networks. After introducing O-RAN and SatCom fundamentals, it examines: (i) NTN architecture compatibility, (ii) virtualization and disaggregation, and (iii) near‑/real‑time control loops. Three integration architectures: Full‑gNB, Orbit‑split and Feeder‑split, are proposed to meet satellite mobility and latency constraints. A detailed study of eight functional split options shows that lower‑layer splits (Options 7, 6) struggle with high throughput and tight latency, whereas higher‑layer splits relax these limits but reduce control granularity; adaptive techniques are outlined to ease interface capacity and delay bottlenecks. Virtualized RAN and CN platforms simplify adoption of new 3GPP NTN features across Releases 17–19, enabling software‑only upgrades of gNBs, UEs and cores. Control‑loop research spans non‑RT, near‑RT and RT operations. A deep‑Q‑learning offloading strategy dynamically balances TN/NTN traffic, boosting spectral efficiency. A neural‑network CSI predictor counters channel aging from satellite motion, improving SCMA decoding accuracy. Finally, an AI‑driven CNF orchestration scheme employs multivariate LSTM forecasting to allocate resources proactively, reducing delay and energy use. Collectively, these contributions show that O-RAN principles, virtualization and AI can overcome NTN‑specific constraints and enable agile, sustainable and high‑performance 6G networks.