Smart Solutions Towards Cellular-Connected Unmanned Aerial Vehicles System (AUTONOMY)

蜂窝连接无人机系统（AUTONOMY）的智能解决方案

• 批准号: EP/W004348/1
• 负责人: Yansha Deng
• 金额: $ 55.11万
• 依托单位: King's College London
• 依托单位国家: 英国
• 项目类别: Research Grant
• 财政年份: 2022
• 资助国家: 英国
• 起止时间: 2022 至 无数据
• 项目状态: 未结题
• 来源: https://gtr.ukri.org/projects?ref=EP%2FW004348%2F1
• 关键词: Smart; Solutions; Towards; Cellular; Connected

The 5G-and-Beyond cellular networks promise UAVs with ultra-reliable low-latency control, ubiquitous coverage, and seamless swarm connectivity under complex and highly flexible multi-UAV behaviours in three-dimension (3D), which will unlock the full potential of UAVs. This so-called cellular-connected UAVs (C-UAVs) system creates a radically different and rapidly evolving networking and control environment compared to conventional terrestrial networks:1) The UAV-ground BS/user channels enjoy fewer channel variations due to their dominant line-of-sight (LOS) characteristics, which imposes severe air-ground interference to the coexisting BSs/users in the uplink/downlink.2) Operating in existing cellular networks designed mainly for dominate downlink traffic (e.g., video), the UAVs with high data rate requirement in uplink payload uploading, and ultra-reliable low-latency communication (URLLC) requirement in downlink command and control communication can hardly be satisfied.3) Maintaining seamless connectivity for mission-centric UAV swarms with 3D high mobility is essential for UAV cooperation but extremely challenging. 4) Controlling a swarm of UAVs to accomplish complex tasks with limited human supervision under the connectivity constraints is of capital importance but challenging.The above challenges can hardly be solved via conventional model-driven approaches, which are limited to performance evaluation or optimisation at one time instant in an offline or semi-offline manner, relying on given ideal probabilistic channel models without time correlation. Meanwhile, the future cellular networks in 5G-and-Beyond moves towards an open, programmable, and virtualised architecture with unprecedented data availability. Both facts mandate a fundamental change in the way we model, design, control, and optimise the C-UAVs system, from reactive/incident driven decoupled networking and control operation to proactive/ data-driven joint network and control design.This project has the ambitious vision to develop artificial intelligence (AI)-powered C-UAVs system with full network automation and conditional control automation, that allow for joint design and optimization of the network operation and the UAVs control in real-time with minimum human supervision and the target of mission completion under the long-term quality of service (QoS) guarantees. The project will engage with the end-users to exploit the C-UAVs applications in surveillance and emergency services in urban areas. Our results on network automation and control automation will directly benefit the telecom manufacturers (e.g., Ericsson AB, Toshiba Europe, AccelerComm), and broader UAV industries (e.g., Airborne Robotics, Thales, Northrop Grumman) internationally with foreseeable industrial impact. The NGMN and CommNet will facilitate the dissemination of the research outcomes nationally and internationally. The development, implementation, and testing of our proposed solutions serve as a platform towards the commercialisation of our research outcomes, putting the UK at the forefront of the "connected aerial vehicles" revolution.

5G及以后的蜂窝网络为无人机提供了超可靠的低延迟控制，无处不在的覆盖范围以及在三维（3D）中复杂和高度灵活的多无人机行为下的无缝群连接，这将释放无人机的全部潜力。与传统的地面网络相比，这种所谓的蜂窝连接的UAV（C-UAV）系统创建了完全不同的并且快速发展的联网和控制环境：1）UAV-地面BS/用户信道由于其主导视线（LOS）特性而享有较少的信道变化，这对上行链路/下行链路中共存的BS/用户施加了严重的空中-地面干扰。视频）、上行载荷上传的高数据速率要求和下行命令和控制通信的超可靠低延迟通信（URLLC）要求的无人机难以满足。3）保持以任务为中心的具有3D高机动性的无人机群的无缝连接对于无人机协作至关重要，但也极具挑战性。4)在连通性约束条件下，控制无人机群在有限的人工监督下完成复杂任务具有重要意义，但也具有挑战性。传统的模型驱动方法仅限于在某一时刻以离线或半离线方式进行性能评估或优化，依赖于给定的理想概率信道模型，没有时间相关性，很难解决上述挑战。与此同时，5G及以后的未来蜂窝网络将朝着开放、可编程和虚拟化的架构发展，具有前所未有的数据可用性。这两个事实都要求我们对C-UAV系统的建模、设计、控制和优化方式进行根本性的改变，从被动/事件驱动的解耦网络和控制操作到主动/数据驱动的联合网络和控制设计。该项目的雄心勃勃的愿景是开发人工智能（AI）驱动的C-UAV系统，具有完全的网络自动化和条件控制自动化，这允许在最少的人工监督和在长期服务质量（QoS）保证下完成使命的目标的情况下实时地对网络操作和UAV控制进行联合设计和优化。该项目将与最终用户合作，利用C-UAV在城市地区的监控和应急服务中的应用。我们在网络自动化和控制自动化方面的成果将直接使电信制造商受益（例如，爱立信公司、东芝欧洲公司、加速通信公司）以及更广泛的无人机行业（例如，机载机器人公司、泰勒斯公司、诺斯罗普·格鲁曼公司）在国际上具有可预见的工业影响。NGMN和CommNet将促进研究成果在国内和国际上的传播。我们提出的解决方案的开发，实施和测试作为我们研究成果商业化的平台，使英国处于“连接飞行器”革命的最前沿。

项目成果

Environment-Aware AUV Trajectory Design and Resource Management for Multi-Tier Underwater Computing

• DOI: 10.1109/jsac.2022.3227103
• 发表时间: 2023-02-01
• 期刊: IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS
• 影响因子: 16.4
• 作者: Hou, Xiangwang;Wang, Jingjing;Hanzo, Lajos
• 通讯作者: Hanzo, Lajos

QoE Optimization for Live Video Streaming in UAV-to-UAV Communications via Deep Reinforcement Learning

• DOI: 10.1109/tvt.2022.3152146
• 发表时间: 2022-05-01
• 期刊: IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY
• 影响因子: 6.8
• 作者: Burhanuddin, Liyana Adilla Binti;Liu, Xiaonan;Zahemszky, Andras
• 通讯作者: Zahemszky, Andras

Non-episodic and Heterogeneous Environment in Distributed Multi-agent Reinforcement Learning

• DOI: 10.1109/globecom48099.2022.10000672
• 发表时间: 2022-12
• 期刊: GLOBECOM 2022 - 2022 IEEE Global Communications Conference
• 作者: Fenghe Hu;Yansha Deng;H. Aghvami

Federated Learning and Meta Learning: Approaches, Applications, and Directions

• DOI: 10.1109/comst.2023.3330910
• 发表时间: 2022-10
• 期刊: IEEE Communications Surveys & Tutorials
• 影响因子: 35.6
• 作者: Xiaonan Liu;Yansha Deng;Arumugam Nallanathan;M. Bennis

Tiled-DASH VR Video Streaming: Design and Implementation

• DOI: 10.1109/iceccme55909.2022.9988103
• 发表时间: 2022-11
• 期刊: 2022 International Conference on Electrical, Computer, Communications and Mechatronics Engineering (ICECCME)
• 作者: Fenghe Hu;Hui Zhou;Yansha Deng;A. Nallanathan;H. Aghvami