面向无人机视频流传输的多路径调度算法

曹昌龙; 李领治; 施连敏; 赵庆越

doi:10.11999/JEIT260002

面向无人机视频流传输的多路径调度算法

doi: 10.11999/JEIT260002 cstr: 32379.14.JEIT260002

曹昌龙¹,
李领治^{2, 3, ,},
施连敏⁴,
赵庆越¹

1.
苏州大学计算机科学与技术学院苏州 215008
2.
苏州大学未来科学与工程学院苏州 215222
3.
通用人工智能与大模型江苏省高校重点实验室苏州 215222
4.
苏州析羽信息科技有限公司苏州 215164

基金项目: 国家自然科学基金(62072321)，国家重点研发计划项目(2023YFB4503100)，江苏省科技计划项目(BZ2024062)，江苏省高等学校自然科学研究项目(22KJA520007)，苏州市科技计划项目(2023SS03，SYG2024144, SYG2025129, SNG2025010)

详细信息

作者简介:
曹昌龙：男，硕士生，研究方向为多路径传输协议、强化学习

李领治：男，副教授，研究方向为智能物联网、多智能体协同

施连敏：男，高级工程师，研究方向为物联网与嵌入式人工智能

赵庆越：男，硕士生，研究方向为无人机路径规划

通讯作者:
李领治　lilingzhi@suda.edu.cn

中图分类号: TN915.04
计量
- 文章访问数: 18
- HTML全文浏览量: 5
- PDF下载量: 3
- 被引次数: 0
出版历程
- 修回日期: 2026-04-09
- 录用日期: 2026-04-09
- 网络出版日期: 2026-04-30

Multipath Scheduling Algorithm for UAV Video Streaming

1.
School of Computer Science and Technology, Soochow University, Suzhou 215008, China
2.
School of Future Science and Engineering, Soochow University, Suzhou 215222, China
3.
Key Laboratory of General Artificial Intelligence and Large Models in Provincial Universities, Soochow University, Suzhou 215222, China
4.
Suzhou Xiyu Information Technology Co., Ltd, Suzhou 215164, China

Funds: The National Natural Science Foundation of China (62072321), The National Key R&D Program of China (2023YFB4503100), The Science and Technology Program of Jiangsu Province (BZ2024062), The Natural Science Foundation of the Jiangsu Higher Education Institutions of China (22KJA520007), The Suzhou Planning Project of Science and Technology (2023ss03, SYG2024144, SYG2025129, SNG2025010)

摘要

摘要: 在无人机实时视频流传输场景中，可利用多路径传输协议的带宽聚合优势提升视频体验质量(QoE)。针对该协议在无人机网络环境下面临的动态异构挑战，该文提出一种基于NeuralUCB算法的多路径调度框架—NeuroFly。首先，将多路径流量调度建模为上下文多臂老虎机(CMAB)问题。然后，基于NeuralUCB设计在线学习策略：融合多维度异构特征构建上下文空间，引入帧优先级驱动的冗余传输机制构建动作空间，并结合多目标奖励函数设计，实现动态异构网络下的自适应流量调度。此外，还设计一种上下文监控机制，能够实时检测上下文分布变化并适时重启学习过程，以提升NeuroFly对环境突变的应对能力。最后，仿真和野外环境下的实验结果表明，与现有先进调度算法和传输方案相比，NeuroFly第99百分位延迟降低19.9～51.0%，并在多项视频QoE指标上取得显著领先：平均视频帧率提升最高达24.6%，图像结构相似性提升最高达49.2%，缓冲时间占比减少13.4～77.6%，证明其在无人机动态异构网络下具有更为出色的持续传输优化能力，能够有效降低传输延迟并提升视频QoE。
- 多路径传输协议 /
- 流量调度 /
- 上下文多臂老虎机 /
- 无人机网络 /
- 视频QoE
Abstract: Objective With the rapid development of the low-altitude economy, Unmanned Aerial Vehicle (UAV) technologies have been widely adopted in scenarios such as emergency rescue, disaster monitoring, and urban security. In these applications, achieving stable, low-latency, and high-fidelity video feedback is critical to mission success. Multipath transport protocols can leverage bandwidth aggregation to improve video Quality of Experience (QoE), thereby providing strong support for UAV video streaming. However, under dynamic and heterogeneous network conditions, the actual performance of multipath transport protocols is highly dependent on the design of multipath scheduling algorithms. To address the challenges posed by dynamic and heterogeneous networks, a variety of scheduling algorithms have been proposed. Heuristic-based schedulers employ carefully designed rules to mitigate head-of-line blocking and inter-path load imbalance to some extent, but their reliance on predefined strategies limits adaptability in highly dynamic environments. Learning-based schedulers, in contrast, continuously learn the mapping between network conditions and scheduling rewards from real-time feedback, enabling adaptive performance optimization. Nevertheless, most existing learning-based schedulers are designed for general network scenarios and are not specifically optimized for UAV networks, and their effectiveness in guaranteeing video QoE remains insufficiently validated. Therefore, there is a pressing need for a multipath scheduling algorithm tailored to UAV video streaming scenarios to fully exploit the performance potential of multipath transport protocols. Methods To address the dynamic and heterogeneous challenges faced by multipath transport protocols in UAV video streaming scenarios, this paper proposes NeuroFly, a multipath scheduling framework based on the NeuralUCB algorithm. In NeuroFly, multipath traffic scheduling is formulated as a Contextual Multi-Armed Bandit (CMAB) problem. A context space is constructed by integrating path state information, video encoding features, and UAV mobility parameters to accurately characterize the transmission environment. In the action space, a frame-priority-driven redundancy transmission mechanism is introduced, where video frames are assigned different transmission priorities according to decoding dependencies, and differentiated redundancy strategies are applied to improve the probability of successful frame delivery. Furthermore, a multi-objective reward function is designed to guide the learning of optimal scheduling policies, enabling adaptive optimization under dynamic and heterogeneous network conditions. In addition, to cope with abrupt environmental changes caused by high UAV mobility, a context monitoring mechanism is integrated into NeuroFly to detect network variations and trigger a two-stage restart strategy. Specifically, a soft restart is activated when gradual context drift is detected to remove outdated historical experience, while a hard restart is performed upon abrupt changes by clearing the replay buffer and reinitializing model parameters to restart learning under a new distribution. Results and Discussions The proposed NeuroFly framework is extensively evaluated in both simulation and real-world environments. First, Mininet-WiFi is employed to simulate realistic UAV network environments to evaluate the overall video QoE performance. The results (Fig. 4) indicate that, compared with state-of-the-art heuristic and learning-based schedulers, NeuroFly achieves comprehensive improvements by fully utilizing aggregated multipath bandwidth. Specifically, the 99th-percentile latency is reduced by 19.9-51.0%, the average video frame rate is increased by up to 24.6%, image structural similarity is improved by up to 49.2%, and buffering time ratio is reduced by 13.4-77.6%, demonstrating its superior capability in guaranteeing video QoE. In addition, real-world experiments (Fig. 6) further confirm that NeuroFly delivers favorable video QoE optimization compared to mature solutions already deployed at scale in production environments in real UAV operational scenarios, demonstrating strong practical applicability and holding promise for large-scale deployment across diverse UAV operation scenarios in the future. Conclusions This paper addresses the key challenges of dynamicity, heterogeneity, and high time variability faced by multipath transport protocols in UAV video streaming scenarios, and proposes an intelligent multipath scheduling framework based on the NeuralUCB algorithm, termed NeuroFly. In this framework, the multipath traffic scheduling problem is formulated as a CMAB problem. By carefully designing the context space, action space, and a multi-objective reward function, online learning and adaptive optimization of traffic allocation policies are achieved. In addition, to further enhance robustness under drastic environmental variations, a lightweight context monitoring mechanism is introduced to continuously detect context distribution drift and restart the learning process when necessary, thereby improving adaptability to abrupt environmental changes. Finally, systematic evaluations are conducted on both simulation platforms and real-world UAV operational environments to comprehensively assess the effectiveness of the proposed approach. Simulation results demonstrate that, compared with state-of-the-art heuristic and learning-based schedulers, NeuroFly achieves consistent improvements across video QoE metrics. Real-world experimental results further indicate that, in real UAV operational scenarios, NeuroFly continues to provide favorable video QoE guarantees when compared with mature solutions that have been widely and long-term deployed in industrial practice. These results collectively validate the practicality, robustness, and engineering viability of NeuroFly, suggesting strong potential for large-scale deployment in UAV applications that are highly sensitive to real-time video quality, such as emergency response, power inspection, agricultural monitoring, and logistics delivery.
- Multipath transport protocol /
- Traffic scheduling /
- Contextual Multi-Armed Bandit (CMAB) /
- UAV networks /
- Video Quality of Experience (QoE)

HTML全文

图 1 NeuroFly系统架构

下载: 全尺寸图片幻灯片

图 2 多变量漂移检测示意图

下载: 全尺寸图片幻灯片

图 3 网络拓扑

下载: 全尺寸图片幻灯片

图 4 仿真环境下的实时视频流传输评估

下载: 全尺寸图片幻灯片

图 5 实验硬件平台

下载: 全尺寸图片幻灯片

图 6 野外环境下的实时视频流传输评估

下载: 全尺寸图片幻灯片

1 NeuroFly多路径调度算法

输入：调度周期持续时间$ {T}_{\text{S}} $，冗余粒度参数$ K $
输出：冗余率决策
(1) 初始化：神经网络参数$ {\theta }_{0} $，经验回放池$ M $
(2) for 时间步$ t=1,\cdots ,T $ do
(3) 　获取当前上下文观测$ \{{\boldsymbol{x}}_{t,{{a}_{s}}}\}_{s=0}^{K} $
(4) 　for each$ a\in \mathcal{A}=\{{a}_{0},{a}_{1},\cdots ,{a}_{K}\} $do
(5) 　　计算动作$ a $的置信上界$ U_{t}^{a} $
(6) 　　令动作$ {a}_{t}=\arg {\max }_{a\in \mathcal{A}}U_{t}^{a} $
(7) 　end for
(8) 　执行动作$ {a}_{t} $，开始冗余传输
(9) 　在调度周期结束时刻计算奖励$ {r}_{t} $
(10) 将样本$ \left\langle {\boldsymbol{x}}_{t,{{a}_{t}}},{r}_{t}\right\rangle $存入经验回放池$ M $
(11) 从$ M $中随机采样一个批次的历史经验
(12) 通过随机梯度下降更新网络参数$ {\theta }_{t} $
(13) end for

下载: 导出CSV

表 2 基础路径配置

	RTT(ms)	抖动(ms)	丢包率(%)	带宽(Mbps)
Path 1	25-50	0-10	0-3	20-30
Path 2	50-100	0-20	0-3	20-30

下载: 导出CSV

参考文献(30)

[1]	钱志鸿, 王义君. 低空经济赋能者: 智能无人机技术体系综述与展望[J]. 电子与信息学报, 2026, 48(1): 1–33. doi: 10.11999/JEIT251246. QIAN Zhihong and WANG Yijun. Intelligent unmanned aerial vehicles for low-altitude economy: A review of the technology framework and future prospects[J]. Journal of Electronics & Information Technology, 2026, 48(1): 1–33. doi: 10.11999/JEIT251246.
[2]	SHEN Dian, YANG Bin, ZHANG Junxue, et al. eMPTCP: A framework to fully extend multipath TCP[J]. IEEE/ACM Transactions on Networking, 2024, 32(6): 5459–5474. doi: 10.1109/TNET.2024.3469396.
[3]	KIMURA B, FERLIN S, PAIVA T, et al. Evaluating adaptive video streaming over multipath QUIC with shared bottleneck detection[J]. ACM Transactions on Multimedia Computing, Communications, and Applications, 2025, 21(9): 246. doi: 10.1145/3711862.
[4]	RAICIU C, PAASCH C, BARRE S, et al. How hard can it be? Designing and implementing a deployable multipath TCP[C]. Proceedings of the 9th USENIX Conference on Networked Systems Design and Implementation, San Jose, USA, 2012: 29.
[5]	PAASCH C, FERLIN S, ALAY O, et al. Experimental evaluation of multipath TCP schedulers[C]. Proceedings of the 2014 ACM SIGCOMM Workshop on Capacity Sharing Workshop, Chicago, USA, 2014: 27–32. doi: 10.1145/2630088.2631977.
[6]	LIM Y S, NAHUM E M, TOWSLEY D, et al. ECF: An MPTCP path scheduler to manage heterogeneous paths[C]. Proceedings of the 13th International Conference on Emerging Networking Experiments and Technologies, Incheon, Republic of Korea, 2017: 147–159. doi: 10.1145/3143361.3143376.
[7]	FERLIN S, ALAY Ö, MEHANI O, et al. BLEST: Blocking estimation-based MPTCP scheduler for heterogeneous networks[C]. 2016 IFIP Networking Conference and Workshops, Vienna, Austria, 2016: 431–439. doi: 10.1109/IFIPNetworking.2016.7497206.
[8]	ZHANG Han, LI Wenzhong, GAO Shaohua, et al. ReLeS: A neural adaptive multipath scheduler based on deep reinforcement learning[C]. IEEE INFOCOM 2019 – IEEE Conference on Computer Communications, Paris, France, 2019: 1648–1656. doi: 10.1109/INFOCOM.2019.8737649.
[9]	WU Hongjia, ALAY Ö, BRUNSTRÖM A, et al. Peekaboo: Learning-based multipath scheduling for dynamic heterogeneous environments[J]. IEEE Journal on Selected Areas in Communications, 2020, 38(10): 2295–2310. doi: 10.1109/JSAC.2020.3000365.
[10]	YANG Wenjun, CAI Lin, SHU Shengjie, et al. QoS-driven contextual MAB for MPQUIC supporting video streaming in mobile networks[J]. IEEE Transactions on Mobile Computing, 2025, 24(4): 3274–3287. doi: 10.1109/TMC.2024.3507051.
[11]	ZHOU Dongruo, LI Lihong, and GU Quanquan. Neural contextual bandits with UCB-based exploration[C]. Proceedings of the 37th International Conference on Machine Learning, Vienna, Austria, 2020: 11492–11502. <b>(查阅网上资料, 未找到对应的出版地信息, 请确认补充)</b>.
[12]	LU T, PÁL D, and PAL M. Contextual multi-armed bandits[C]. Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics, Sardinia, Italy, 2010: 485–492.
[13]	HUA Boyu, NI Haoran, ZHU Qiuming, et al. Channel modeling for UAV-to-ground communications with posture variation and fuselage scattering effect[J]. IEEE Transactions on Communications, 2023, 71(5): 3103–3116. doi: 10.1109/TCOMM.2023.3255900.
[14]	LI Lihong, CHU Wei, LANGFORD J, et al. A contextual-bandit approach to personalized news article recommendation[C]. Proceedings of the 19th International Conference on World Wide Web, Raleigh, USA, 2010: 661–670. doi: 10.1145/1772690.1772758.
[15]	VALKO M, KORDA N, MUNOS R, et al. Finite-time analysis of kernelised contextual bandits[C]. Proceedings of the Twenty-Ninth Conference on Uncertainty in Artificial Intelligence, Bellevue, USA, 2013: 654–663.
[16]	FILIPPI S, CAPPÉ O, GARIVIER A, et al. Parametric bandits: The generalized linear case[C]. Proceedings of the 24th International Conference on Neural Information Processing Systems, Vancouver, Canada, 2010: 586–594.
[17]	BIFET A and GAVALDÀ R. Learning from time-changing data with adaptive windowing[C]. Proceedings of the 2007 SIAM International Conference on Data Mining (SDM), Minneapolis, USA, 2007: 443–448. doi: 10.1137/1.9781611972771.42.
[18]	SEEMANN M. quic-go[EB/OL]. https://github.com/quic-go/quic-go, 2025. (查阅网上资料,年份信息不确定,请确认).
[19]	LIU Yanmei, MA Yunfei, DE CONINCK Q, et al. Multipath extension for QUIC[EB/OL]. https://www.ietf.org/archive/id/draft-ietf-quic-multipath-10.html, 2024.
[20]	DOS REIS FONTES R, ROTHENBERG C E. Mininet-WiFi: A platform for hybrid physical-virtual software-defined wireless networking research[C]. Proceedings of the 2016 ACM SIGCOMM Conference, Florianopolis, Brazil, 2016: 607–608. doi: 10.1145/2934872.2959070.
[21]	ROOSENDAAL T. Big buck bunny[C]. ACM SIGGRAPH ASIA 2008 Computer Animation Festival, Singapore, Singapore, 2008: 62.
[22]	NI Yunzhe, ZHENG Zhilong, LIN Xianshang, et al. CellFusion: Multipath vehicle-to-cloud video streaming with network coding in the wild[C]. Proceedings of the ACM SIGCOMM 2023 Conference, New York, USA, 2023: 668–683. doi: 10.1145/3603269.3604832.
[23]	FROMMGEN A, ERBSHÄUSSER T, BUCHMANN A, et al. ReMP TCP: Low latency multipath TCP[C]. 2016 IEEE International Conference on Communications (ICC), Kuala Lumpur, Malaysia, 2016: 1–7. doi: 10.1109/ICC.2016.7510787.
[24]	XING Yitao, XUE Kaiping, ZHANG Yuan, et al. A stream-aware MPQUIC scheduler for HTTP traffic in mobile networks[J]. IEEE Transactions on Wireless Communications, 2023, 22(4): 2775–2788. doi: 10.1109/TWC.2022.3213638.
[25]	E Jinlong, HE Lin, ZHAO Zongyi, et al. AggDeliv: Aggregating multiple wireless links for efficient mobile live video delivery[C]. IEEE INFOCOM 2024 – IEEE Conference on Computer Communications, Vancouver, Canada, 2024: 1173–1180. doi: 10.1109/INFOCOM52122.2024.10621184.
[26]	XING Yitao, XUE Kaiping, ZHANG Yuan, et al. A low-latency MPTCP scheduler for live video streaming in mobile networks[J]. IEEE Transactions on Wireless Communications, 2021, 20(11): 7230–7242. doi: 10.1109/TWC.2021.3081498.
[27]	YANG Wenjun, CAI Lin, SHU Shengjie, et al. Scheduler design for mobility-aware multipath QUIC[C]. GLOBECOM 2022–2022 IEEE Global Communications Conference, Rio de Janeiro, Brazil, 2022: 2849–2854. doi: 10.1109/GLOBECOM48099.2022.10001247.
[28]	XING Yitao, XUE Kaiping, ZHANG Yuan, et al. An online learning assisted packet scheduler for MPTCP in mobile networks[J]. IEEE/ACM Transactions on Networking, 2023, 31(5): 2297–2312. doi: 10.1109/TNET.2023.3246168.
[29]	HAN Xueqiang, HAN Biao, LI Ruidong, et al. MARS: An adaptive multi-agent DRL-based scheduler for multipath QUIC in dynamic networks[C]. 2023 IEEE/ACM 31st International Symposium on Quality of Service (IWQoS), Orlando, USA, 2023: 1–10. doi: 10.1109/IWQoS57198.2023.10188744.
[30]	HAN Jiangping, XUE Kaiping, LI Jian, et al. EdAR: An experience-driven multipath scheduler for seamless handoff in mobile networks[J]. IEEE Transactions on Wireless Communications, 2023, 22(10): 6839–6852. doi: 10.1109/TWC.2023.3246082.