文章背景与核心概要
本文提出了一种分层的开放式无线接入网(O-RAN)控制框架,用于管理运行在频率范围 2(FR2/毫米波)频谱上的无人机(UAV)搭载的 5G 新空口基站(gNB)。由于无人机的移动性改变物理信道特性、阻塞和用户覆盖的速度远快于用户业务需求的变化,传统的孤立优化方法往往难以奏效。
为了解决这种缓慢的空中拓扑管理与快速资源分配之间的耦合问题,作者提出了一种利用 RAN 智能控制器(RIC)的多时间尺度架构:非实时 RIC(rApp)联合控制系留无人机的位置以及增强型移动宽带(eMBB)和超高可靠低时延通信(URLLC)的切片预算;近实时 RIC(xApp)则实现为一种排列等变的 DeepSets Soft Actor-Critic (D-SAC) 调度器,将用户视为无序集合,在 rApp 的切片预算内分配每个用户的资源。实验结果表明,该分层控制器在 eMBB SLA 满足率上提升高达 17%,在 URLLC 按时交付率上提升高达 42%。
Place, Slice and Schedule: Hierarchical O-RAN Control of a Tethered mmWave UAV-gNB
Authors: Alireza Mohammadhosseini, Fatemeh Afghah
Primary Subject: Networking and Internet Architecture (cs.NI)
Secondary Subject: Artificial Intelligence (cs.AI)
Submitted: August 24, 2026 (arXiv:2608.23824 [cs.NI])
Full-Text Links: View PDF | TeX Source
📌 Summary
本文提出了一种分层的开放式无线接入网(O-RAN)控制框架,用于管理运行在频率范围 2(FR2/毫米波)频谱上的无人机(UAV)搭载的 5G 新空口基站(gNB)。由于无人机的移动性改变物理信道特性、阻塞和用户覆盖的速度远快于用户业务需求的变化,传统的孤立优化方法往往难以奏效。
This paper presents a hierarchical Open Radio Access Network (O-RAN) control framework for managing unmanned aerial vehicle (UAV)-mounted 5G New Radio base stations (gNBs) operating in the Frequency Range 2 (FR2/mmWave) spectrum. Because UAV mobility dynamically alters physical channel characteristics, blockages, and user coverage at a much faster rate than user traffic demands evolve, traditional isolated optimizations fall short.
为了解决这种缓慢的空中拓扑管理与快速资源分配之间的耦合问题,作者提出了一种利用 RAN 智能控制器(RIC)的多时间尺度架构: - 非实时 RIC (rApp): 利用聚合的 KPI 和无线电环境上下文联合控制系留无人机的位置以及增强型移动宽带(eMBB)和超高可靠低时延通信(URLLC)的切片预算。 - 近实时 RIC (xApp): 实现为一种排列等变的 DeepSets Soft Actor-Critic (D-SAC) 调度器,将用户视为无序集合,在 rApp 的切片预算内分配每个用户的资源。
To resolve this coupling between slow aerial topology management and fast resource allocation, the authors propose a multi-timescale architecture utilizing a RAN Intelligent Controller (RIC): - Non-Real-Time RIC (rApp): Uses aggregated KPIs and radio-environment context to jointly control tethered UAV placement and the slice budget for enhanced Mobile Broadband (eMBB) and Ultra-Reliable Low-Latency Communication (URLLC). - Near-Real-Time RIC (xApp): Implemented as a permutation-equivariant DeepSets Soft Actor-Critic (D-SAC) scheduler that treats users as an unordered set, allocating per-user resources within the rApp's slice budget.
通过使用 Sionna RT 射线追踪信道进行训练,与经典基线和学习型基线相比,该分层控制器实现了高达 17% 的 eMBB SLA 满足率提升 以及高达 42% 的 URLLC 按时交付率提升。此外,经过训练的 rApp 还使 URLLC 的按时交付率额外提升了 20%。
Trained using Sionna RT ray-traced channels, the hierarchical controller achieves up to a 17% improvement in eMBB SLA satisfaction and up to a 42% improvement in URLLC on-time delivery over classical and learned baselines. Furthermore, the learned rApp provides an additional 20% increase in URLLC on-time delivery.
📑 Metadata & Reference
- Cite as:
Mohammadhosseini, A., & Afghah, F. (2026). Place, Slice and Schedule: Hierarchical O-RAN Control of a Tethered mmWave UAV-gNB. arXiv:2608.23824. - DOI: 10.48550/arXiv.2608.23824
- Comments: Submitted to IEEE for possible publication.
- Cite as:
Mohammadhosseini, A., & Afghah, F. (2026). Place, Slice and Schedule: Hierarchical O-RAN Control of a Tethered mmWave UAV-gNB. arXiv:2608.23824.- DOI: 10.48550/arXiv.2608.23824
- Comments: Submitted to IEEE for possible publication.