Abstract: Short, discrete inter-layer visibility windows, link setup time, and dynamic task arrivals complicate remote-sensing data return through a low-Earth-orbit (LEO) relay layer. An event-driven rolling beam search method is proposed. The scheduling model considers visibility windows, link setup time, task deadlines and priorities, and relay access constraints. At each event, executable connections are retained; candidate connections are evaluated by task priority, transmission gain, and opportunity-preservation cost, and explored over subsequent event states. For dynamic arrivals, arrival distributions estimate future resource demand and adjust candidate evaluation. When all tasks were known a priori but had different deadlines and priorities, the task, priority-weighted, and high-priority on-time completion rates were 82.22%, 77.65%, and 73.61%. Under dynamic arrivals, distribution information increased the task and high-priority on-time completion rates and the average effective data-return completion ratio by 3.9, 12.7, and 1.2 percentage points, respectively. The method balances current data-return gains against future access opportunities under limited future information.

Key words: remote sensing data return, inter-layer links, resource scheduling, visibility constraints, link establishment delay, LEO relay constellation