中图分类号:TP393文献标志码:ADOI:10.19358/j.issn.2097-1788.2026.01.004 中文引用格式:张策,苏思雨. 基于改进YOLOv8n的通信终端识别算法[J].网络安全与数据治理,2026,45(1):20-28. 英文引用格式:Zhang Ce,Su Siyu. Emitter identification algorithm for communication terminals based on improved YOLOv8n[J].Cyber Security and Data Governance,2026,45(1):20-28.
Emitter identification algorithm for communication terminals based on improved YOLOv8n
Zhang Ce1,Su Siyu2
1. Jiayuan Science and Technology Co.,Ltd.; 2. School of Communication Engnineering, Hangzhou Dianzi University
Abstract: To address the issue of decreased identification accuracy of communication terminals caused by signal interference in complex electromagnetic environments, an improved YOLOv8n-based emitter identification algorithm for communication terminals is proposed, named EMI-YOLO. Firstly, to tackle the problem of interference signals occluding the target signal, a C2fCE module is proposed, which integrates deep convolution, pointwise convolution, and the Efficient Channel Attention (ECA) mechanism to expand the model′s receptive field. Secondly, a partial selfattention mechanism is embedded at the end of the backbone network to enhance the model′s ability to learn signal features. Furthermore, five data augmentation strategies are employed to effectively expand the dataset. The experimental results indicate that EMIYOLO demonstrates a 7.4% improvement in mAP5095 than YOLOv8n in the training set, with a reduction of 0.4M in model parameters; compared to three baseline algorithms, EMIYOLO improves the identification accuracy for six mobile phone models by an average of 42.3%, 52%, 53.4%, 50.4%, 34% and 39.7% in the test set, respectively. Therefore, EMIYOLO exhibits strong antiinterference capability and robustness in complex electromagnetic environments.
Key words : communication terminal identification; YOLOv8n; complex electromagnetic environment