Deploying Intelligence Solutions into Non-Terrestrial Network Based O-RAN Architectures

Non-Terrestrial Networks (NTNs) are expected to play a central role in future 6G systems by complementing terrestrial infrastructures with wide-area coverage, resilience, and service continuity. In parallel, Open Radio Access Network (O-RAN) architectures introduce functional disaggregation, open interfaces, and multi-time-scale control mechanisms that enable flexible and intelligent network operation. Integrating O-RAN principles into NTN environments, however, raises fundamental challenges related to latency, resource constraints, and the feasibility of distributed control and Machine Learning (ML)–driven optimization. This paper investigates the deployment of intelligence solutions in NTN-based O-RAN architectures from a system-level and network-design perspective. We analyze how RAN functions and control entities can be distributed across space–ground infrastructures and identify representative deployment options for Radio Unit, Distributed Unit, and Central Unit functions. The interaction between O-RAN control loops operating at different time scales and the constraints imposed by NTN environments is examined, with particular attention to the placement of ML training and inference functionalities. Rather than proposing a specific optimization algorithm, the paper structures the design space of Space O-RAN deployments by providing a comparative analysis of architectural trade-offs and an analytical feasibility assessment of control-loop execution under realistic NTN latency conditions. The results highlight which configurations can support near-real-time control and latency-sensitive services, and which are limited to long-term optimization. The proposed framework offers practical design insights for the development of scalable and intelligent NTN-enabled O-RAN systems.