Possible Controller Placement of Mega-Constellated LEO Satellite Network: Cornerstone Toward Satellite Network for AI

This study explores the placement of satellite controllers to ensure efficient and reliable communication in mega-constellation Low Earth Orbit (LEO) satellite networks. Effective controller or primary node placement is essential not only for managing dynamic satellite networks but also for offloading data and enabling AI functionalities within these networks. After qualitatively identifying the intra-constellation scenario as a feasible architecture among various options, this study primarily proposes a systematic tiling-based controller placement paradigm and rigorously evaluates its structural resiliency and latency performance. A tiling-based controller placement model is proposed for LEO satellites at the same altitude, aiming to satisfy the satellite requirements outlined in IMT-2020. To evaluate network stability and reliability, the Breadth-First Search algorithm is used to calculate the path and hop count from each satellite node to its nearest controller. Performance is assessed by analyzing the average hop count under different failure scenarios, including single and double inter-satellite link (ISL) failures, satellite node failures, and controller satellite failures. Additionally, the study examines the trend in average hop count as the failure rate of satellites or ISLs increases. Latency of control signals is analyzed for specific scenarios. Although this study focuses on the controller placement problem, the proposed framework can be extended to other applications, such as satellite data centers.