无线电工程2025，Vol.55Issue(3)：633-641,9.DOI:10.3969/j.issn.1003-3106.2025.03.021
无人机网络的用户关联与神经网络模型云边协同剪枝算法
User Association in UAV Networks and Cloud-edge Collaborative Pruning Algorithm Based on Neural Networks
杨健 1陶金玉 2王成 2杨清海2
作者信息
- 1. 中国电子科技集团公司第五十四研究所,河北 石家庄 050081
- 2. 西安电子科技大学 通信工程学院,陕西 西安 710071
- 折叠
摘要
Abstract
In order to solve the problems of limited UAV storage and insufficient overall network performance in the field of UAV network and edge intelligence,a cloud-edge collaborative pruning algorithm based on reinforcement learning user association strategy and neural network model is proposed.An association algorithm between UAV users and base stations is proposed,which optimizes the UAV network topology based on reinforcement learning.According to the output of the user association algorithm,the cloud and edge of the network are divided,and the allocation type of the neural network model and the size requirements of the model are planned.Furthermore,a cloud-edge collaborative pruning algorithm is proposed.By fully utilizing the local data of individual UAV,the neural network model of UAV network collaboration is streamlined,communication consumption is reduced,and the overall performance of the UAV network is further improved.The results show that the user association strategy based on reinforcement learning improves the channel capacity.Compared with the traditional centralized pruning model,the cloud-edge cooperative pruning algorithm consumes lower resources on the uplink communication link.The proposed method achieves efficient data transmission and calculation processing in a UAV collaborative environment,and provides an effective way to solve the challenges of communication and computing integration under resource constraints.关键词
无人机网络/边缘智能/用户关联策略/神经网络模型精简Key words
UAV network/edge intelligence/user association strategy/neural network model compression分类
信息技术与安全科学引用本文复制引用
杨健,陶金玉,王成,杨清海..无人机网络的用户关联与神经网络模型云边协同剪枝算法[J].无线电工程,2025,55(3):633-641,9.
