Synopsis
This paper surveys the application and development of Artificial Intelligence (AI) in Satellite Communication (SatCom) and Non-Terrestrial Networks (NTN). It presents a comprehensive list of use cases, challenges, and the main AI tools capable of addressing these challenges. The paper also discusses the advantages and disadvantages of onboard and on-ground AI-based architectures and reviews current commercial and research activities in this field.
Importance of AI/ML in SATCOM
The Satellite Communication ecosystem is experiencing a revolution with the emergence of lower orbits as a low-latency alternative to conventional Geo-Stationary Orbit communication systems. This new era brings fundamental operational challenges, such as the need for autonomously adaptive mechanisms, which are difficult to manage through human intervention. AI promises to solve these challenges by enabling automation and data-driven techniques, potentially reducing operational expenditure significantly.
Machine-Learning Assisted Satellite System
The satellite communications world is progressing towards integrating Machine Learning (ML) for specific use cases. The digitalization of satellite payloads enhances flexibility, allowing for efficient support of varying communication demands. ML aids in optimizing complex procedures, predicting network loads, and assisting in dynamic resource allocation. Additionally, the evolution of the ground segment towards a multi-gateway environment necessitates ML-based automation for optimizing network operations.
Onboard or On-ground ML Dilemma
A critical dilemma in satellite communication is whether to perform operations onboard the satellite or on the ground. Onboard strategies increase complexity and resource requirements, while on-ground methods alleviate onboard complexity but can introduce delays. The industry is developing AI-specific processors to enable efficient execution of complex algorithms in both scenarios.
Ongoing Activities and Development
The paper overviews significant activities by private and public institutions in developing ML in satellite communication. Projects like MIRSAT testbed and SPIRE’s Brain in Space project focus on experimenting with new autonomous network algorithms. Agencies like ESA and NASA are also actively deploying ML technologies in space for various applications.
Use Cases Analysis
The paper details various satellite communication use cases, grouped into three layers that align with the conventional OSI layers. For each use case, the paper presents the motivation, detailed description, conventional solutions, and ML solutions, providing insight into the practical applications and benefits of ML in satellite communications.
Summary
The paper provides a comprehensive overview of the opportunities and challenges presented by AI and ML in improving the performance and efficiency of NTNs. It underscores the transformative impact of these technologies on satellite communications, addressing current trends, operational challenges, and future research directions.