LEO-Aware DRL Meta-Scheduler for 5G Non-Terrestrial Network Slicing

2026-08-03

AI Summary

This work addresses the challenge of heterogeneous network slicing resource management in low Earth orbit non-terrestrial networks (LEO NTN) integrated into 5G/6G systems, where high propagation delays, rapid satellite mobility, and non-stationary channels complicate service provisioning. To tackle this, the authors propose a dual-timescale scheduling mechanism based on a disaggregated Open RAN architecture: a slow meta-scheduler selects slice policies every 100 ms using stale telemetry data, while a fast MAC scheduler handles user requests at each transmission time interval (TTI), leveraging Twin Delayed Deep Deterministic Policy Gradient (TD3) reinforcement learning to model orbital dynamics and service-level agreement (SLA) constraints. The approach uniquely achieves stringent control over both variance and magnitude of queuing delay for mission-critical traffic in LEO NTN slicing, while preventing resource starvation in broadband slices caused by conventional max-CQI strategies, thereby ensuring robust inter-slice isolation. Simulations demonstrate that the scheme meets strict RLC-layer queuing delay requirements for mission-critical services across varying loads, sacrificing only a statistically insignificant 1% of system capacity (p > 0.05).

Abstract

The integration of Low Earth Orbit (LEO) Non-Terrestrial Networks (NTNs) into 5G and upcoming 6G architectures introduces various challenges, including severe propagation delays, ultra-high base station mobility, and channel non-stationarity, complicating radio resource management of heterogeneous network slices. In this paper, we propose a deep reinforcement learning (DRL) meta-scheduler for twin-timescale resource allocation. Our solution adopts a decoupled Open Radio Access Network (RAN) architecture, in which a strategic 100 ms meta-scheduler selects scheduling policies for the different network slices using stale telemetry, while a fast-timescale MAC packet scheduler processes per-TTI user requests. The resulting Markov Decision Process captures non-stationary orbital dynamics and heterogeneous SLAs constraints via a TD3 agent. Simulation results under varying traffic load show that, unlike other solutions, the proposed meta-scheduler explicitly trades a statistically insignificant 1% capacity fraction (p > 0.05) to strictly bound the variance and overall magnitude of RLC-layer queuing delay for Mission-Critical (MC) traffic. Crucially, it enforces this isolation without inducing the broadband slice starvation characteristic of standard maximum-CQI heuristics, establishing a robust foundation for 6G O-RAN NTN resource allocation.

Problem

• LEO NTN
• network slicing
• radio resource management
• non-stationary channel
• propagation delay

Innovation

• LEO NTN
• DRL meta-scheduler
• network slicing
• twin-timescale scheduling
• O-RAN