탐욕 알고리즘을 이용한 단일 저궤도 위성의 다중 타겟 관측 스케줄링

초록

In this paper, we perform observation scheduling for a single Low Earth Orbit (LEO) satellite with a task of observing multiple targets on Earth, aiming for a schedule with minimal speed change. For optimal scheduling, Genetic and Greedy Algorithms (GA and GrA, for short) are introduced and their solutions are compared and analyzed. GA spends a relatively large amount of computational cost but has a merit of returning a near-optimal solution. In contrast, GrA spends a little amount of computational cost but yields a solution with a large distance to the optimal. Therefore, GrA is applied with a variable maximum-observation-time-interval to yield a solution whose quality is comparable to that of GA. In fact, it is found that the GrA solution’s quality approaches the GA one for the observation time interval of 1.9 day, in case of visiting 10 targets. This result and its aspect of computational complexity are thoroughly verified and confirmed on random sets of target positional data with various numbers of targets. Also, the effect of satellite’s roll maneuver is investigated and a new data visualization scheme (so-called Snail Plot) is introduced to effectively explain the concept of GrA.

키워드