The objectives of the activity are to investigate the applicability of ML or AI techniques in in the overall domain of satellite communications. The activity shall further prototype and demonstrate a limited number of Proof of Concepts (PoCs). These PoCs shall lead to follow-up activities in other ARTES Elements.
The activity will identify multiple use cases in the domain of satellite communications that could be addressed by using ML/AI techniques. Such use cases could either be existing problems or new concepts that could enable new capabilities for satellite communication systems. The use case could address any phase in the lifecycle of satellite communication systems.
As a minimum, some use cases shall be investigated that address the reduction of interference, the optimisation of spectrum usage and radio resources in scenarios where satellite systems interfere with each other, or in which satellite systems interfere with terrestrial systems. Other use cases shall could be of interest is optimisation of network management and operations for large complex satellite constellations.
The activity shall evaluate a number use cases and ML/AI techniques, including supervised and unsupervised learning, deep and continuous learning and constrained based learning. The activity shall actively seek the feedback from industry on the practical applicability of identified techniques. Based on justified trade-off criteria and the consultations with industry, a subset of the identified use cases shall be simulated or emulated to make a convincing demonstration of the added value of such ML/AI techniques. The activity shall identify additional activities which shall be initiated.
The activity will generate the following outcomes:
- An assessment of the applicability of AI/ML techniques in satellite communications in general
- A number of proof-of-concepts that will demonstrate the added value of AI/ML
- A roadmap for further developments in the area ofAI/ML