Resource Slicing with Cross-Cell Coordination in Satellite-Terrestrial Integrated Networks

TL;DR

The paper tackles resource slicing in satellite-terrestrial integrated networks (STIN) under spatiotemporal demand dynamics and satellite mobility. It introduces a hybrid data-model co-driven distributed resource slicing (DRS) framework that combines an asynchronous multi-agent proximal policy optimization (AMAPPO) for satellite selection with a distributed optimization-based reservation scheme, enabling scalable cross-cell coordination. The problem is formulated as a long-term optimization (P0) and decomposed into a short-window reservation subproblem (P1) and a long-term satellite selection subproblem (P2); AMAPPO with CTDE guides satellite choices, while a convex, penalty-enabled reservation algorithm (IDOA) coordinates resource allocation within slicing windows. Simulation results show faster convergence and competitive cost-delay performance compared to baselines, demonstrating the approach’s potential for practical, delay-aware, cross-cell resource management in STIN. The work advances scalable, low-latency STIN resource slicing and points to future enhancements in beam collaboration and adaptive beam resource sharing.

Abstract

Satellite-terrestrial integrated networks (STIN) are envisioned as a promising architecture for ubiquitous network connections to support diversified services. In this paper, we propose a novel resource slicing scheme with cross-cell coordination in STIN to satisfy distinct service delay requirements and efficient resource usage. To address the challenges posed by spatiotemporal dynamics in service demands and satellite mobility, we formulate the resource slicing problem into a long-term optimization problem and propose a distributed resource slicing (DRS) scheme for scalable and flexible resource management across different cells. Specifically, a hybrid data-model co-driven approach is developed, including an asynchronous multi-agent reinforcement learning-based algorithm to determine the optimal satellite set serving each cell and a distributed optimization-based algorithm to make the resource reservation decisions for each slice. Simulation results demonstrate that the proposed scheme outperforms benchmark methods in terms of resource usage and delay performance.

Figures (4)

• Figure 1: Resource slicing in STIN.
• Figure 2: Convergence performance.
• Figure 3: Elevation angles impacts.
• Figure 4: CDF of system cost in each time slot.