无人机辅助动态权重边缘计算卸载策略研究

王义君; 王雅出; SHAHDBatool; 缪瑞新

doi:10.11999/JEIT260054

无人机辅助动态权重边缘计算卸载策略研究

doi: 10.11999/JEIT260054 cstr: 32379.14.JEIT260054

长春理工大学电子信息工程学院长春 130022

基金项目: 国家自然科学基金(61771219)，吉林省自然科学基金(20250102227JC).

详细信息

作者简介:
王义君：男，教授，研究方向为通感一体无线网络通信及面向极端环境与特殊场景通信

王雅出：女，硕士研究生，研究方向为无人机边缘计算

SHAHDBatool：女，硕士研究生，研究方向为智能反射面资源分配

缪瑞新：男，讲师，研究方向为核磁共振成像及物联网技术

通讯作者:
王义君　wyjs-107@163.com

中图分类号: TN929.5
计量
- 文章访问数: 19
- HTML全文浏览量: 3
- PDF下载量: 1
- 被引次数: 0
出版历程
- 修回日期: 2026-03-24
- 录用日期: 2026-03-24
- 网络出版日期: 2026-04-22

Research on UAV-Assisted Dynamic Weighted Edge Computing Offloading Strategy

School of Electronics and Information Engineering, Changchun University of Science and Technology, Changchun 130022, China

Funds: The National Natural Science Foundation of China (61771219), Jilin Provincial Natural Science Foundation (20250102227JC).

摘要

摘要: 针对无人机辅助移动边缘计算环境下计算资源受限、系统处理任务总开销过高问题，提出基于协同缓存自适应的分层多元宇宙优化(CCAH-MVO)算法优化卸载策略。首先，构建微云－边缘－本地三层网络架构，在无人机边缘服务器上预制缓存程序，采用细粒度部分卸载策略，并针对多无人机覆盖的终端设备制定无人机选择策略。然后，提出CCAH-MVO算法协同优化缓存、卸载和资源分配，并引入动态权重机制自适应平衡时延与能耗，得到最优卸载策略。仿真结果表明,所提策略的时延更优、能耗在裕度区内可控，综合性能优于基准卸载策略。
- 移动边缘计算 /
- 卸载决策 /
- 无人机选择策略 /
- 动态权重 /
- 多元宇宙
Abstract: Objective As the Internet of Things (IoT) continues to evolve with increasing demands for computational resources and real-time processing, the significance of Mobile Edge Computing (MEC) technology has become increasingly prominent. However, traditional MEC heavily relies on terrestrial base stations, resulting in coverage blind spots in remote or specialized environments. Unmanned Aerial Vehicle (UAV)-assisted MEC architectures leverage the flexible deployment capabilities of UAVs to effectively expand service coverage. Nonetheless, in complex scenarios involving multiple user devices and UAV-based servers, existing research remains insufficient in simultaneously optimizing task offloading latency, system energy consumption, and adaptability to dynamic environments. Specifically, current approaches often overlook the optimal server selection problem when terminal devices are covered by multiple UAVs and lack mechanisms for adaptive adjustment of optimization objectives based on system states during task processing. This paper aims to address the challenges of collaborative caching, offloading decision-making, and resource allocation in multi-terminal, multi-UAV scenarios. By introducing a dynamic weighting mechanism and advanced optimization algorithms, the proposed framework seeks to achieve low-latency performance while effectively managing system energy consumption, ultimately reducing total system overhead and enhancing user experience. Methods This paper constructs a "micro-cloud-edge-terminal" three-tier collaborative computing architecture consisting of a central cloud, multiple UAV edge servers with caching capabilities, and numerous mobile terminal devices. In this model, a cache-assisted mechanism is introduced to reduce transmission delay during task execution. Task offloading adopts a fine-grained partial offloading mode, which divides complex tasks into subtasks with dependencies and models them through a directed acyclic graph (DAG). To address the core optimization problem, a Cooperative Caching-Adaptive Hierarchical Multiverse Optimizer (CCAH-MVO) algorithm is proposed. Firstly, a hybrid coding scheme is designed to uniformly encode offloading decisions, caching decisions, and resource allocation. Then, a dynamic weight mechanism is introduced to adaptively adjust the weight ratio of delay and energy consumption in the total overhead according to the system energy consumption state. Additionally, a UAV selection strategy for multi-UAV coverage scenarios is proposed. By simulating inter-universe material exchange and local refined search, the optimal offloading strategy is efficiently found. This method is implemented on the MATLAB platform and validated through extensive simulations under various experimental settings. Results and Discussions To verify the performance of the proposed strategy, 50 randomly distributed terminal devices and 5 UAVs are set up in a 400m×400m simulation scenario. Comparisons are conducted with algorithms such as SCA, PSO-GWO, and I-ACO, as well as strategies including local execution, full offloading, and fixed weights. The initial simulation state shows that the 5 UAVs are deployed above the centers of terminal device clusters, and terminal devices located at the edges of the clusters are simultaneously within the communication coverage of multiple UAVs (Fig. 5). In this scenario, the optimal UAV is selected by calculating the UAV selection function value (Fig. 6), which effectively avoids resource bottlenecks caused by all terminals flocking to a single UAV and achieves balanced load distribution. In terms of delay performance, the CCAH-MVO algorithm maintains the lowest delay across the entire task quantity range, with a gentle increase as the number of tasks grows (Fig. 7). Among different offloading strategies, the delay of the CCAH-MVO algorithm's offloading strategy is lower than that of the fixed-weight strategy throughout the task quantity range, demonstrating the superiority of the dynamic adaptive mechanism in maintaining low delay (Fig. 10). In terms of energy consumption performance, the energy consumption differences among various algorithms are small when the task quantity is low. Under high task load, the energy consumption curve tends to flatten after the dynamic weight mechanism is activated (Fig. 8). When the number of tasks reaches 100, the total energy consumption of the CCAH-MVO strategy is the lowest among all strategies and lower than that of the fixed-weight strategy, reflecting the effective control capability of the dynamic weight mechanism in the critical energy consumption state (Fig. 9). In terms of total system overhead, the CCAH-MVO algorithm consistently maintains the optimal performance, and the gap with the fixed-weight strategy gradually expands when the number of tasks exceeds 80, demonstrating the collaborative optimization capability of dynamic weights for delay and energy consumption (Fig. 11). Overall, by integrating the dynamic weight mechanism and achieving load balancing through the UAV selection strategy, the CCAH-MVO algorithm effectively solves the problems of resource constraints and excessively high total system task processing overhead in complex and dynamic UAV-assisted mobile edge computing environments, realizing precise coordination between delay and energy consumption at different load stages. Conclusions Aiming at the task offloading problem in multi-UAV-assisted mobile edge computing scenarios, this paper proposes a collaborative optimization strategy based on the improved CCAH-MVO algorithm. By constructing a three-tier network architecture, introducing a caching mechanism, adopting fine-grained offloading, and designing dynamic weight and UAV selection strategies, the strategy effectively solves the resource scheduling dilemma in complex environments. Comparative simulation experiments with various strategies confirm that the proposed strategy can adaptively adjust optimization objectives according to the real-time system state, intelligently control energy consumption while ensuring low latency, significantly reduce total system overhead, and improve service stability and user experience. This research provides a valuable solution for the efficient management of UAV edge computing resources in dynamic environments. Future work will further deepen the exploration of dynamic energy efficiency optimization and multi-node collaboration mechanisms while maintaining the advantage of low latency.
- Mobile Edge Computing (MEC) /
- Offloading Decision-Making /
- UAV Selection Strategy /
- Dynamic Weight /
- Multi-Verse Optimizer Algorithm (MVO)

HTML全文

图 1 系统模型图

下载: 全尺寸图片幻灯片

图 2 有向无环图

下载: 全尺寸图片幻灯片

图 3 初始开销矢量空间图

下载: 全尺寸图片幻灯片

图 4 混合编码结构图

下载: 全尺寸图片幻灯片

图 5 多无人机终端覆盖分布图

下载: 全尺寸图片幻灯片

图 6 无人机选择函数值对比图

下载: 全尺寸图片幻灯片

图 7 任务数量变化下不同算法时延性能的比较

下载: 全尺寸图片幻灯片

图 8 任务数量变化下不同算法能耗性能的比较

下载: 全尺寸图片幻灯片

图 9 多种策略的任务卸载总能耗

下载: 全尺寸图片幻灯片

图 10 多种策略的任务卸载总时延

下载: 全尺寸图片幻灯片

图 11 多种策略的任务卸载总开销

下载: 全尺寸图片幻灯片

1 协同缓存自适应的分层多元宇宙优化算法

输入：无人机集合$ \boldsymbol{N} $、终端设备集合$ \boldsymbol{M} $、任务集合$ \boldsymbol{L} $、程序集　合$ \boldsymbol{J} $、系统参数($ {Q}_{n} $,$ F_{n}^{l} $,$ F_{m}^{l} $,$ B $等)
输出：最优卸载策略(卸载决策$ \boldsymbol{A} $、缓存策略$ \boldsymbol{C} $、资源分配方案　$ \boldsymbol{R} $)
1. 初始化参数：迭代次数$ {T}_{\max } $、种群规模$ {P}_{\text{size}} $、裕度区阈值　$ \delta =0.8 $；
2. 采用K-means++聚类确定无人机悬停位置；
3. FOR每个宇宙$ i $：
4. 　生成混合编码向量$ {\boldsymbol{V}}_{i} $，满足缓存约束；
5. 　计算个体适应度值$ {H}_{L}\left({\boldsymbol{V}}_{i}\right)=\displaystyle\sum \nolimits_{i=1}^{k}{H}_{l} $，得到最优宇宙；
6. END FOR
7. WHILE $ t\leq {T}_{\max } $：
8. 　更新宇宙膨胀率；
9. 　根据$ {T}_{l}/{T}_{\text{loc}}/\sqrt{{\left({T}_{\mathrm{l}}/{T}_{\text{loc}}\right)}^{2}+{\left({E}_{\mathrm{l}}/{E}_{\text{loc}}\right)}^{2}} $调整$ \beta $；
10. 计算选择矩阵$ \boldsymbol{S} $；
11. FOR每个宇宙$ i $：
12. 　根据式(38)、式(39)、式(40)更新$ a_{k,\mathrm{n}}^{\text{new}} $、$ c_{j,\mathrm{n}}^{\text{new}} $、$ b_{m}^{\text{new}} $，更　　　新当前宇宙；
13. 　修复缓存容量和任务依赖；
14. END FOR
15. FOR每个宇宙$ i $：
16. 　重新计算更新后适应度值；
17. END FOR
18. $ t=t+1 $；
19. END WHILE

下载: 导出CSV

表 1 仿真参数

仿真参数	参考数值
无人机数量N	5
终端设备数量M	50
信道总带宽B	20 MHz
终端设备传输功率$ {P}_{m} $	0.2 W
噪声功率密度$ {N}_{0} $	–168 dBm/Hz
参考信道增益$ {h}_{0} $	–60 dB
无线回程链路容量$ {R}_{\mathrm{b}} $	100 Mbps
覆盖角$ \theta $	$ \pi /6 $
每个子任务数据量$ D_{m}^{\text{in}} $	$ \left[0.2{,}0.4\right] $ MB
任务所需CPU周期数$ {W}_{m} $	$ \left[1.0{,}1.6\right] $ Giga-cycles
终端设备计算能力$ F_{m}^{l} $	0.5 GHz
无人机计算能力$ F_{n}^{l} $	$ \left[10{,}15\right] $ GHz
无人机飞行速度$ \upsilon $	15 m/s
无人机传输功率$ {P}_{u} $	0.6 W
能耗系数κ	$ 1.0\times {10}^{-27} $
无人机存储容量$ {Q}_{n} $	300 MB

下载: 导出CSV

参考文献(25)

[1]	PHAM Q V, FANG Fang, HA V N, et al. A survey of multi-access edge computing in 5G and beyond: Fundamentals, technology integration, and state-of-the-art[J]. IEEE Access, 2020, 8: 116974–117017. doi: 10.1109/ACCESS.2020.3001277.
[2]	HUANG Xiaoyao, ZHANG Baoxian, and LI Cheng. Incentive mechanisms for mobile edge computing: Present and future directions[J]. IEEE Network, 2022, 36(6): 199–205. doi: 10.1109/MNET.107.2100652.
[3]	王涵, 文方青, 李兴旺, 等. 大规模MIMO OTFS系统二维稀疏恢复信道估计方法研究[J]. 通信学报, 2025, 46(8): 165–175. doi: 10.11959/j.issn.1000-436x.2025146. WANG Han, WEN Fangqing, LI Xingwang, et al. Two-dimensional sparse recovery method for channel estimation in massive MIMO OTFS systems[J]. Journal on Communications, 2025, 46(8): 165–175. doi: 10.11959/j.issn.1000-436x.2025146.
[4]	李兴旺, 田志发, 张建华, 等. IRS辅助NOMA网络下隐蔽通信性能研究[J]. 中国科学: 信息科学, 2024, 54(6): 1502–1515. doi: 10.1360/SSI-2023-0174. LI Xingwang, TIAN Zhifa, ZHANG Jianhua, et al. Performance analysis of covert communication in IRS-assisted NOMA networks[J]. Scientia Sinica Informationis, 2024, 54(6): 1502–1515. doi: 10.1360/SSI-2023-0174.
[5]	王义君, 李嘉欣, 闫志颖, 等. 基于深度强化学习的移动边缘计算安全传输策略研究[J]. 通信学报, 2025, 46(4): 272–281. doi: 10.11959/j.issn.1000-436x.2025060. WANG Yijun, LI Jiaxin, YAN Zhiying, et al. Research on secure transport strategy of mobile edge computing based on deep reinforcement learning[J]. Journal on Communications, 2025, 46(4): 272–281. doi: 10.11959/j.issn.1000-436x.2025060.
[6]	TANG Chaogang, DING Yao, XIAO Shuo, et al. Collaborative service caching, task offloading, and resource allocation in caching-assisted mobile edge computing[J]. IEEE Transactions on Services Computing, 2025, 18(4): 1966–1981. doi: 10.1109/TSC.2025.3586093.
[7]	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.
[8]	DENG Yiqin, ZHANG Haixia, CHEN Xianhao, et al. UAV-assisted MEC with an expandable computing resource pool: Rethinking the UAV deployment[J]. IEEE Wireless Communications, 2024, 31(5): 110–116. doi: 10.1109/MWC.015.2300427.
[9]	SUN Lu, LIU Ziqian, NING Zhaolong, et al. Multi-agent q-net enhanced coevolutionary algorithm for resource allocation in emergency human-machine fusion UAV-MEC system[J]. IEEE Transactions on Automation Science and Engineering, 2025, 22: 4473–4489. doi: 10.1109/TASE.2024.3409551.
[10]	BASHARAT M, NAEEM M, KHATTAK A M, et al. Digital-twin-assisted task offloading in UAV-MEC networks with energy harvesting for IoT devices[J]. IEEE Internet of Things Journal, 2024, 11(23): 37550–37561. doi: 10.1109/JIOT.2024.3440061.
[11]	ZHOU Fuhui, WU Yongpeng, HU R Q, et al. Computation rate maximization in UAV-enabled wireless-powered mobile-edge computing systems[J]. IEEE Journal on Selected Areas in Communications, 2018, 36(9): 1927–1941. doi: 10.1109/JSAC.2018.2864426.
[12]	LIU Boyang, WAN Yiyao, ZHOU Fuhui, et al. Resource allocation and trajectory design for MISO UAV-assisted MEC networks[J]. IEEE Transactions on Vehicular Technology, 2022, 71(5): 4933–4948. doi: 10.1109/TVT.2022.3140833.
[13]	PENG Chaoda, HUANG Xumin, WU Yuan, et al. Constrained multi-objective optimization for UAV-enabled mobile edge computing: Offloading optimization and path planning[J]. IEEE Wireless Communications Letters, 2022, 11(4): 861–865. doi: 10.1109/LWC.2022.3149007.
[14]	YU Sicong, ZHENG Huiji, and MA Caihong. MEC-enabled fine-grained task offloading for UAV networks in urban environments[J]. Sustainability, 2022, 14(21): 13809. doi: 10.3390/su142113809.
[15]	LI Hui, LI Xiuhua, FAN Qilin, et al. Distributed DNN inference with fine-grained model partitioning in mobile edge computing networks[J]. IEEE Transactions on Mobile Computing, 2024, 23(10): 9060–9074. doi: 10.1109/TMC.2024.3357874.
[16]	CHEN Xing, LI Ming, ZHONG Hao, et al. FUNOff: Offloading applications at function granularity for mobile edge computing[J]. IEEE Transactions on Mobile Computing, 2024, 23(2): 1717–1734. doi: 10.1109/TMC.2023.3240741.
[17]	DENG Min, YAO Zhiqiang, LI Xingwang, et al. Dynamic multi-objective AWPSO in DT-assisted UAV cooperative task assignment[J]. IEEE Journal on Selected Areas in Communications, 2023, 41(11): 3444–3460. doi: 10.1109/JSAC.2023.3310056.
[18]	SACCO A, ESPOSITO F, MARCHETTO G, et al. A self-learning strategy for task offloading in UAV networks[J]. IEEE Transactions on Vehicular Technology, 2022, 71(4): 4301–4311. doi: 10.1109/TVT.2022.3144654.
[19]	CHEN Haosheng, GUI Haixia, WANG Jiahuan, et al. Computation offloading optimization for UAV-based cloud-edge collaborative task scheduling strategy[J]. IEEE Transactions on Cognitive Communications and Networking, 2025, 11(6): 4240–4253. doi: 10.1109/TCCN.2025.3544822.
[20]	ALYASSI R, KHONJI M, KARAPETYAN A, et al. Autonomous recharging and flight mission planning for battery-operated autonomous drones[J]. IEEE Transactions on Automation Science and Engineering, 2023, 20(2): 1034–1046. doi: 10.1109/TASE.2022.3175565.
[21]	王亭惠, 陈桂芬. 基于DP-HAFS算法的移动边缘计算卸载策略[J]. 计算机应用研究, 2023, 40(4): 1184–1188. doi: 10.19734/j.issn.1001-3695.2022.08.0419. WANG Tinghui and CHEN Guifen. Mobile edge computing offloading strategy based on DP-HAFS algorithm[J]. Application Research of Computers, 2023, 40(4): 1184–1188. doi: 10.19734/j.issn.1001-3695.2022.08.0419.
[22]	LI Ze. Enhanced fault localization in energy systems using an improved MVO algorithm and multistrategy optimization[J]. IEEE Access, 2025, 13: 50367–50378. doi: 10.1109/ACCESS.2025.3552761.
[23]	LI Zhiyang, CHEN Ming, CHEN Jinli, et al. Delay efficient caching enabled hierarchical mobile edge computing networks[J]. IEEE Transactions on Communications, 2025, 73(10): 9087–9101. doi: 10.1109/TCOMM.2025.3562526.
[24]	GUL F, RAHIMAN W, ALHADY S S N, et al. Meta-heuristic approach for solving multi-objective path planning for autonomous guided robot using PSO–GWO optimization algorithm with evolutionary programming[J]. Journal of Ambient Intelligence and Humanized Computing, 2021, 12(7): 7873–7890. doi: 10.1007/s12652-020-02514-w.
[25]	SARI D W, DWIJAYANTI S, and SUPRAPTO B Y. Integration of regression-based guidance ant for enhanced exploration and convergence in ant colony optimization (ACO)[J]. IEEE Access, 2025, 13: 107621–107630. doi: 10.1109/ACCESS.2025.3581425.