Satellite Constellation Design Method for Long-Term Simulations Using Spherical Nearest Neighbor Index

In this paper, we address the consideration of designing satellite constellations for Earth and celestial bodies like the Moon and Mars, where the influence of complex perturbations complicates orbit prediction. Traditionally, long-term simulations have been employed to calculate performance metrics such as coverage, resulting in a significant computational load due to the large amount of data generated. A common approach to reduce this has been to compute performance over one orbital period and extrapolate it to estimate overall performance. However, for celestial bodies with locally irregular gravity fields like the Moon and Mars, this method can distort actual orbital periodicity and fail to account for perturbations, leading to reduced accuracy in coverage calculations. To overcome these limitations, we propose a method that improves the accuracy of performance metric calculations at each point, utilizing an alternative metric based on the Spherical Nearest Neighbor Index (SNNI) instead of conventional coverage metrics. Comparing optimization results using conventional metrics with those based on the proposed SNNI-based alternative, we demonstrate a significant reduction in computation time.