Optimal design of sparse satellite coverage constellations using genetic algorithms

Abstract

This thesis is mainly concerned with the problem of designing target orbits for sparse coverage constellation mission using a few low Earth orbit (LEO) satellites. In particular, a temporary reconnaissance mission design problem is introduced, and a new approach for this type of problem is proposed in this study. It is difficult to solve this type of problem by an analytical approach or traditional gradient-based optimization methods because each LEO satellite is currently in its own orbit with limited fuel budget for orbital maneuvering, and even the sensor characteristics of each satellite are not identical. Therefore, the genetic algorithm is applied to overcome discontinuity, non-differential nature, and many local minima of the problem. However, the performance of genetic algorithms tends to be problem-dependant. Thus, various efforts to improve the fitness function in dealing with the genetic operators and parameters satisfying the fuel constraints are performed, and their effects on the performance of the algorithm are discussed. Throughout a comprehensive simulation study, it is successfully demonstrated that a new approach based on the genetic algorithm could provide preliminary target orbits as an initial starting point to form a temporary constellation satisfying the constraint (i.e., the fuel limitations). Consequently, the guideline of applying the genetic algorithm for this type of problem is empirically presented. The proposed approach can be applied to other types of sparse coverage problems consisting of multiple tactical satellites that are more affordable under a specified fuel limitation and are capable of achieving multiple mission objectives.