Space Edge Computing Solution

CEOS2030, a Space Edge Computing solution for Earth observation satellites

The CEOS2030 project, labeled by Aerospace Valley and supported by France 2030, represents a major step forward for onboard intelligence in space.

Led by a consortium headed by STEEL Electronique and bringing together ELSYS Design (an ADVANS Group subsidiary), IP-Maker, Tarides, and Thales Alenia Space, with support from IRT Saint Exupéry, CEOS2030 aims to integrate onboard computing capabilities into Earth observation satellites.

The objective is to process data as close as possible to its source in order to optimize resources, reduce unnecessary data flows, and improve responsiveness to environmental events.

Presentation of CEOS2030, a Space Edge Computing solution for Earth observation satellites.

Why Space Edge Computing is essential for Earth observation

Earth observation satellites are critical to understanding and protecting our environment: biodiversity monitoring, forest surveillance, water resource management, and natural disaster detection.

Today, most data is downlinked to the ground before analysis, leading to delays and significant resource consumption. Up to 70% of the images transmitted by space missions are never exploited.

CEOS2030 introduces a paradigm shift: extracting value from data onboard and transmitting only what is essential. This approach frees up resources for other uses and optimizes satellite capabilities.

An onboard data center for Space Edge Computing

To address the challenges of Earth observation, CEOS2030 delivers a genuine technological breakthrough: integrating an “onboard data center” directly into satellites. This approach is based on an open and scalable architecture capable of hosting new applications throughout the mission, representing a major advance compared to today’s fixed systems.

At the core of the solution are space-grade, radiation-hardened electronic components, including next-generation FPGAs such as AMD Xilinx Versal and Kalray processors. This strategic choice provides an excellent balance between performance and power consumption, while ensuring robustness against radiation, requirements that conventional GPUs cannot meet in the space domain.

Built on this hardware layer, CEOS2030 integrates advanced data processing mechanisms: embedded artificial intelligence algorithms capable of detecting clouds, forest fires, or environmental changes in real time. These processing capabilities drastically reduce the volume of data to be transmitted to the ground while improving responsiveness to critical events.

Application orchestration relies on ORCHIDE, a solution developed by Thales Alenia Space, which uses unikernel containers to ensure lightweight and cyber-secure orchestration. Combined with a minimalist operating system designed by Tarides, this approach provides unprecedented flexibility and enhanced security. NVMe protocols and compression mechanisms further optimize data storage and transmission.

The roadmap is ambitious: a full demonstration is planned for 2027 using representative hardware (TRL 6), followed by in-orbit integration starting in 2028. Ultimately, CEOS2030 could equip future generations of European satellites, offering a sovereign and sustainable solution for onboard space data processing.

“The idea behind ORCHIDE and CEOS2030 is to achieve European autonomy in the field of Space Edge Computing.

With ORCHIDE, we have the application orchestration layer on satellites, and with CEOS2030, we have the onboard hardware to make this project a reality.” (Hervé Didtsch, Head of R&D for Observation and Science Systems at Thales, interviewed in the September 2025 issue of L’Informaticien)

The role of ADVANS Group

Within the CEOS2030 consortium, ELSYS Design and AVISTO, two ADVANS Group subsidiaries, bring complementary expertise to address the challenges of Space Edge Computing.

ELSYS Design focuses on integrating embedded artificial intelligence models at the FPGA level. Its role is to adapt and optimize AI algorithms so they can operate efficiently in the constrained space environment, where power consumption and robustness are critical. This contribution includes porting models to FPGA, hardware validation, and implementing solutions that enable multiple AI models to run in parallel while meeting the operational requirements of future missions.

AVISTO is involved upstream in the software development of AI algorithms. It contributes to model creation, optimization, and the preparation of training datasets. AVISTO also ensures software quality and the structuring of deliverables, guaranteeing that the developed solutions are robust, well-documented, and compliant with space industry standards.

Together, these two ADVANS Group entities help make CEOS2030 a sovereign and scalable solution capable of integrating new applications throughout the satellite lifecycle. This collaboration demonstrates the group’s ability to combine expertise in microelectronics, artificial intelligence, and software engineering to address the strategic challenges of European space programs.

“The core challenge is enabling onboard learning. It is not just about detecting clouds or trucks on a road, but about identifying changes in behavior within a scene. The goal is to go beyond basic image recognition use cases and to embed multiple AI models on the same satellite, in order to address the needs of several customers using a single platform.” (Jean‑François Béraud, interviewed in the September 2025 issue of L’Informaticien)

A long-term ambition

CEOS2030 follows a progressive approach: while technologies evolve rapidly, their integration into space systems requires long cycles (around 10 years on average). This initiative lays the groundwork for future generations of satellites, with a modular and agile vision designed to meet evolving user needs.

Interested in working with us on your next project?
Fill out the contact form and we’ll get back to you.