Edge computing in non-terrestrial networks: architecture and operational constraints
Non-terrestrial network (NTN) architectures integrate low-Earth orbit (LEO) satellite nodes and stratospheric platforms (HAPS) as active segments of the communications chain. In this context, the deployment of processing capacity directly on these nodes — referred to as NTN edge computing — responds to a specific architectural requirement: to reduce the volume of traffic that must be routed to the terrestrial core and to shorten the decision loop in applications with strict latency requirements. The technical premise is that part of the logical processing, data filtering or execution of network functions can be carried out on the airborne or space node itself, without requiring a round trip to the ground segment. The practical implementation of this model presents significant constraints. LEO nodes operate under severe restrictions of mass, energy consumption and thermal dissipation, which limits the onboard computing capacity and necessitates a very precise design of the processing payloads. On HAPS platforms, conditions are somewhat less restrictive in terms of mass and power supply, but prolonged exposure to high-altitude environments introduces reliability and fault-tolerance requirements that condition component selection and the design of management software. In both cases, updating the onboard software — necessary to adapt virtualised network functions — entails telecommand procedures with bounded operational windows. From the perspective of load distribution, NTN architecture with edge computing requires precise definition of which functions reside at the edge and which remain in the core. Functions that are candidates for the edge include the pre-processing of observation data, local mobility management between beams or between nodes, and the execution of quality-of-service (QoS) policies without dependence on signalling towards the ground. However, the consistency of network state across distributed nodes — particularly in constellations with dynamic topology — demands synchronisation and orchestration mechanisms that add complexity to the control plane. The regulatory and standardisation framework relevant to these architectures is developed principally within 3GPP for NTN integration in mobile networks, and within ETSI for virtualised network functions (NFV) applied to non-terrestrial infrastructures. The convergence between these frameworks and the specific requirements of the space segment — in particular the management of the return link and propagation latency — constitutes one of the technical axes around which the design of NTN systems with distributed processing capability is structured. For operators and institutions evaluating these architectures, the critical variable is not the raw computing capacity at the edge, but the functional coherence of the complete system under conditions of degraded link or partial node failure.
NASSAT - Network Satellite Systems