聊城大学学报（自然科学版）2026，Vol.39Issue(2)：224-237,14.DOI:10.19728/j.issn1672-6634.2025040006
基于DB-GS-Yolo11的跨域智能无线感知算法
Intelligent wireless cross-domain sensing algorithm based on DB-GS-Yolo11
摘要
Abstract
Wireless sensing technology utilizes WiFi signals in the environment to extract feature informa-tion and identify target motion states.With the widespread adoption of smart devices,this technology has been extensively applied in fields such as smart homes,healthcare,human-computer interaction,and au-tonomous driving.However,due to the complex and dynamic nature of mobile communication environ-ments,wireless sensing faces challenges such as low model accuracy,poor scenario generalization,and high environmental dependency.To address these issues across diverse cross-domain scenarios,we pro-pose DB-GS-Yolo11,a dual-branch gated sequential Yolo11-based cross-domain intelligent wireless sensing algorithm.The algorithm employs a dual-branch architecture,integrating Yolo11 neural networks,a Ga-ted Attention Coding(GAC)module,and a State Space Model(SSM).This design enables efficient signal perception and robust extraction of key cross-domain features,significantly improving the model's general-ization capability.The proposed enhancement substantially reduces environmental dependency,endowing the system with greater robustness,portability,and cross-domain recognition accuracy.In comparative experiments,the proposed DB-GS-Yolo11 algorithm is compared in performance with various mainstream neural network models,including Deep Neural Network(DNN),Gated Recurrent Unit(GRU),and Google Inception Net neural network(GoogLeNet).The experimental results show that DB-GS-Yolo11 exhibits superior performance in optimizing perceptual complexity,improving recognition speed,and cross domain adaptability.The overall perception accuracy in the domain dataset has improved by 5.33%to 9.67%,and the recognition efficiency has increased by 1.35%to 17.81%.The recognition accuracy of the algorithm proposed on cross domain datasets such as cross position and cross direction has been improved by 2.33%to 7.33%and 2.67%to 4.33%while the recognition efficiency has been improved by 1.63%to 4.69%and 0.23%to 3.11%respectively.关键词
智能无线感知/跨域识别/注意力机制/双支路Yolo11神经网络Key words
intelligent wireless sensing/cross-domain recognition/attention mechanism/dual-branch Yo-lo11 neural network分类
信息技术与安全科学引用本文复制引用
孙海洋,李天成,刘广虎,徐凌伟..基于DB-GS-Yolo11的跨域智能无线感知算法[J].聊城大学学报（自然科学版）,2026,39(2):224-237,14.基金项目
国家自然科学基金项目(62201313) （62201313）
数字化学习技术集成与应用教育部工程研究中心创新基金项目(1321012) （1321012）
广西高校非线性电路与光通信重点实验室开放基金课题(NCOC-25-03) （NCOC-25-03）
国家级大学生创新创业训练计划项目(202410426019)资助 （202410426019）
