Network Slice-Based Low-Altitude Intelligent Network for Advanced Air Mobility
Kai Xiong, Yutong Chen, Zongmiao He, Yujie Qin, Supeng Leng, Chau Yuen
Abstract
Advanced Air Mobility (AAM) represents a transformative approach to urban transportation through electric vertical take-off and landing vehicles (eVTOLs). However, the operational constraints of eVTOLs, including limited onboard computational resources and battery capacity, necessitate efficient task offloading strategies to ground infrastructure. This paper proposes a novel Network Slice (NS)-based framework for task offloading within Low-Altitude Intelligent Networks (LAIN), specifically designed for AAM systems operating in vertically layered airspace. Our approach integrates multi-agent Reinforcement Learning (RL) and Deep Deterministic Policy Gradient (DDPG) techniques to dynamically allocate heterogeneous resources (bandwidth, beam alignment, and computing capacity) according to the diverse requirements of eVTOL tasks. The framework features an access pairing mechanism for optimal eVTOL-BS-slice assignment and a deep RL-based slice orchestration system for resource allocation and lifecycle management. Simulation results demonstrate that our approach significantly outperforms existing benchmarks in resource allocation efficiency and operational cost reduction across varying eVTOL velocities and restricted airspace conditions. This work provides critical insights for designing intelligent network slicing solutions for future AAM transportation ecosystems.