Puncture-Aware Resource Allocation for Maritime Satellite-Terrestrial Networks

The rising demand for reliable maritime wireless communication has driven the development of hybrid satellite-terrestrial networks, where satellites provide wide-area coverage and Terrestrial Base Stations (TBSs) offer high-rate nearshore access. In this context, shipborne users require heterogeneous services, such as enhanced Mobile BroadBand (eMBB) and Ultra-Reliable Low-Latency Communication (URLLC), which impose conflicting Quality of Service (QoS) demands on limited spectrum resources. To address these challenges, we propose a deep reinforcement learning framework for dynamic resource management. The proposed algorithm jointly performs resource allocation—including link selection, bandwidth provisioning, and power control—followed by puncturing scheduling. The puncturing mechanism preempts eMBB transmissions to accommodate URLLC packets with strict latency and reliability requirements, thereby enabling efficient coexistence under limited resources. It operates on two-time scales, assigning eMBB resources at the timeslot level and conducting URLLC puncturing at the mini-slot level. Unlike traditional sequential approaches, our method synchronizes cross-layer decisions, achieving better tradeoffs between URLLC reliability and eMBB throughput. Simulation results verify that the proposed method significantly improves task completion rates while achieving higher efficiency and stability compared to baseline algorithms.