Distributed Solving Approaches at the Edge for Constellation Observations Problems

Increasing developments address Space Situation Awareness and Surveillance for Earth monitoring, global warming and astrophysics. Development of new spacecraft constellation approaches has become affordable for private companies and nations, and where Edge AI can be applied. However, those systems require resources and performance optimisation, but also resilient and reactive behaviours, with no central point of failure while limiting vulnerabilities. We propose a simplified model for solving cooperative Earth Observation Satellite Scheduling (EOSS), with either centralized or distributed solving approaches. Our contribution consists in a method to provide adequate models and distributed algorithms, and to evaluate search techniques in order to fit distributed edge Earth observation. We evaluate the resilience of the constellation for both approaches, while identifying promising domains to be investigated. Preliminary results are presented, that show the interest and the potential of our techniques.