Optimal Continuous-Time Job Scheduling for Multiple Low Earth Orbit Satellites

Abstract

This paper proposes a mixed-integer linear program (MILP) formulation for job scheduling of a constellation of low earth orbit satellites and investigates the applicability/scalability of a standard MILP solver that produces the optimal solution. The goal of the satellite constellation job scheduling is to allocate each job for satellites and to determine the job starting times in order to maximize the overall mission performance measure. The scheduling problem is formulated by first selecting and timetabling the job (observation activities) to acquire the user-requested data of the Earth surface, with incorporating satellite operational constraints such as visibility time windows, transition time between consecutive jobs, maximum attitude angle, energy capacity, memory capacity. The proposed formulation relaxes some of these constraints, which would not have impacts on real instances, but additionally includes precedence condition between jobs and job-agent compatibility constraints. An off-the-shelf MILP solver is used to obtain the optimal solution for this scheduling formulation; numerical experiments designed for investigating the applicability of the optimal solver in terms of problem size indicates that the optimal solution can be obtained in a tractable manner up to the problem size with three satellites and hundreds of jobs.