Towards dynamic real-time scheduling for multiple earth observation satellites

Abstract

Currently, most tasks submitted to earth observation satellites (EOSs) are with deadlines, to satisfy users' timing requirements. Besides, tasks are normally submitted dynamically, with uncertainties of tasks number and submission times. In this paper, we establish a multi-objective mathematic programming model for dynamic real-time scheduling of EOSs. To improve user's revenue and resource utilization, a task dynamic merging strategy is proposed. With the dynamic merging in place, plus consideration of retracting and rearranging some scheduled tasks to make room for incorporating the newly arrived tasks, we present a novel dynamic real-time scheduling algorithm'DMTRH. To demonstrate the superiority of our DMTRH, we conduct extensive experiments by simulations to compare DMTRH with two algorithms'RHTSS and STI. The experimental results indicate that DMTRH improves the scheduling quality of others and is suitable for dynamic real-time scheduling.