Simulated annealing and genetic algorithm approaches for discontinuous coverage satellite constellation design

Constellations of Earth orbiting satellites are used to perform many tasks. If the constellation's coverage need not be continuous in time, the number of satellites and, therefore, the cost of the constellation may be decreased. Finding constellations that minimize revisit times for discontinuous coverage is posed as an optimization problem. Practical limits on coverage analysis provide objective functions that are not differentiable, so non-gradient-based optimization methods are necessary. Simulated annealing and a genetic algorithm were applied to discontinuous coverage problems for various numbers of satellites and Earth central angles, providing both whole Earth and zonal coverage. Results are presented and compared to constellation designs from a traditional approach. Simulated annealing appears to provide answers with less computational effort, but the genetic algorithm is capable of simultaneously generating different designs with the same revisit times. Additional conclusions about these approaches for discontinuous coverage satellite constellation design are made.