• So, you have been tasked to build a LEO communications satellite solution, and you and your team must balance customer needs with constraints such as cost, performance, mass, thermal limits, reliability, lifespan, and time to market. Your team must analyze launch and orbital strategies alongside spacecraft and payload design to determine satellite quantity, placement, and deployment logistics. Key decisions include defining ground terminal requirements, spectrum selection, payload capabilities (availability, throughput, coverage), and backhaul strategies. Critical trade-offs—such as standardized vs. flexible vs. custom payload solutions, power efficiency, ISLs, customer metrics, satellite size, payload field of view, bent-pipe vs. regenerative payloads, and the role of software-defined networks (SDNs)—must be evaluated and prioritized. The evolving landscape of industry standards, declining launch costs, emerging technologies like AI, and shifting market forces must also be considered. Ultimately, the goal for you and your team (if you choose to take on this mission) is to craft a differentiated and cost-effective customer experience within the constraints of investment capital, design for space, customer costs, and a competitive global market landscape.

  • AI Foundation models are reshaping Earth Observation (EO) by replacing raw image distribution with compact semantic embeddings which are high-level representations that power downstream AI systems across diverse sectors. These embeddings are not only valuable on the ground but will increasingly be used onboard satellites to enable fast, efficient communication between space assets and ground stations. This shift raises important questions: Should embeddings be explainable or even human-readable? Or can we embrace more abstract forms that allow AIs to reason with concepts beyond human understanding? Current models rely on a narrow set of training tasks. We introduce a new framework inspired by human cognitive development, where foundation models are progressively trained to acquire spatial, temporal, and causal reasoning abilities. Early results show that this leads to more efficient, generalizable, and interpretable embeddings, paving the way for a new generation of embedding-centric EO infrastructures, powered by AI foundation models to communicate with other AIs
     

  • The scope and speed of innovation in satellite communications have increased extensively in the last years.

    ESA is addressing some of the future trends covering commercial and institutional markets and services. 

    Among them: 

    • TN/NTN integration and convergence around the 3GPP set of standards – 5G and 6G
    • Secure connectivity: IRIS2 and the work toward the backbone connectivity in a space system of systems concept
    • Optical and quantum communications: from classical links, through QKD and towards the IoQT (Internet of the Quantum Things)
    • Moonlight as the stepping stone towards the Solar System Internet