Kenny Andersson, Swedish Defence Research Agency; Tim Kinnunen, Swedish Defence Research Agency; Semeli Papadogiannakis, Swedish Defence Research Agency; Rolf Ragnarsson, Swedish Defence Research Agency
Keywords: Space situational awareness, Sensor planning, Optimization, Radar, Sensor resource allocation
Abstract:
The ability to track, catalogue, monitor, and characterise satellites is critical for maintaining space situational awareness. Radar-based sensors play a key role in this effort, offering all weather, day-and-night capabilities. However, observation planning is complex and efficiently allocating radar resources to track a large number of satellites is a non-trivial task. This paper proposes an observation planning algorithm for allocating sensor resources to measure a population of satellites within a planning duration. The objective is to formulate this as an optimisation problem and implement a genetic algorithm to produce observation schedules. Our approach keeps selection variables binary, ensuring clear and unambiguous solutions. While genetic algorithms may not always converge to the global optimum, they require less computation time, which enables more frequent schedule updates. Additionally, instead of enforcing strict uncertainty requirements, which can waste resources on satellites with already well determined orbits, soft constraints are used that aims to improve overall performance across all satellites, even if it allows for a slight increase in individual uncertainties. The soft constraints, in the form of penalty functions, allow different strategies to be defined. These can be chosen to allow for more aggressive, conservative, or opportunistic scheduling strategies. Three different scheduling methods are simulated to illustrate the optimisation, ranging from purely greedy to more dynamic. The backbone of these methods are the same, with the difference being that the more dynamic methods have the ability to reallocate tasked observations.
Date of Conference: September 16-19, 2025
Track: Machine Learning for SDA Applications