🤖 AI Summary
To address severe cross-layer interference arising from spectrum sharing in vertical heterogeneous networks (vHetNets) integrating High-Altitude Platform Stations (HAPS) and terrestrial macro base stations in urban environments, this paper presents a lightweight interference coordination framework that systematically compares centralized versus distributed interference management paradigms for the first time. The framework synergistically integrates convex optimization, game theory, and distributed reinforcement learning, operating on local information exchange to reduce computational complexity by over 40% and improve network scalability by up to threefold, while maintaining spectral efficiency above 95%. Its core contribution lies in rigorously establishing the intrinsic advantages of distributed AI methods—namely, low overhead and high resilience—in interference management, thereby delivering a deployable, scalable interference mitigation paradigm for large-scale air-ground integrated networks.
📝 Abstract
Next-generation wireless networks are evolving towards architectures that integrate terrestrial and non-terrestrial networks (NTN), unitedly known as vertical heterogeneous networks (vHetNets). This integration is vital to address the increasing demand for coverage, capacity, and new services in urban environments. In vHetNets, various tiers can operate within the same frequency band, creating a harmonized spectrum-integrated network. Although this harmonization significantly enhances spectral efficiency, it also introduces challenges, with interference being a primary concern. This paper investigates vHetNets comprising high altitude platform stations (HAPS) and terrestrial macro base stations (MBSs) operating in a shared spectrum, where interference becomes a critical issue. The unique constraints of HAPS-enabled vHetNets further complicate the interference management problem. To address these challenges, we explore various strategies to manage interference in HAPS-enabled vHetNets. Accordingly, we discuss centralized and distributed approaches that leverage tools based on mathematical optimization and artificial intelligence (AI) to solve interference management problems. Preliminarily numerical evaluations reveal that distributed approaches not only achieve lower complexity but also deliver superior scalability compared to centralized methods, primarily due to their reduced dependence on global information.