Agile Satellite Scheduling via Permutation Search with Constraint Propagation

Earth Observation Satellite (EOS) scheduling is an important oversubscribed constraint optimization problem. Permutation-based scheduling methods have recently been shown to be effective on these problems. However, the new agile EOS satellites present additional scheduling complexity because they allow image acquisition over a window of possible observation times. Constraint propagation algorithms have been successfully applied in traditional local search methods for these problems. In this paper, we describe a synthesis of permutation-based search and constraint propagation for agile EOS scheduling. Our approach incorporates the advantages of both techniques. We obtain the large neighbourhood behaviour of permutation search for oversubscribed resource scheduling problems. As well, we exploit the power of constraint propagation to retain as much flexibility as possible while building the schedule. We investigate different local optimization algorithms (including hill-climbing, simulated annealing and squeaky wheel optimization) coupled with constraint propagation over image acquisition time windows. We compare our method to recent permutation-based methods for non-agile EOS scheduling which rely upon a greedy scheduler for assigning image acquisition times. Experiments are performed on synthetic EOS data sets using both uniform random image targets and actual urban image target sets. We measure both schedule quality and solution degradation as new image requests are added dynamically to the problem. Our results suggest that permutation-based search coupled with constraint propagation works very well for agile EOS scheduling.