Attackers exploit multiple vulnerabilities in the Internet of Things for continuous or concurrent attacks, while traditional detection methods are limited to a single stage and difficult to fully capture attack characteristics and stage changes, resulting in insufficient utilization of feature extraction and residual mapping information, affecting the accuracy of abnormal pattern recognition and signal prediction, and reducing the efficiency of multi-stage attack detection. Therefore, a multi-stage attack anomaly traffic detection algorithm for cellular IoT is proposed. By analyzing the multi-stage attack process of cellular IoT in depth, accurately identify the characteristics of each stage. The obtained cellular IoT operation signals are standardized and converted into image form. The multi-scale mixed residual module is used to extract bee feature images, which improves the accuracy and comprehensiveness of feature extraction. By fitting residual blocks through identity mapping, residual mapping can be more effectively utilized to optimize the performance of multi-stage attack anomaly traffic detection and accurately reflect data anomaly patterns. Finally, utilizing the fully connected layer Softmax function to achieve signal category prediction and efficiently detect multi-stage attack abnormal traffic. Through experimental analysis, when the number of multi-scale mixed modules is 10, the algorithm achieves the best detection effect of abnormal traffic in multi-stage attacks. At this time, the packet loss rate of cellular iot data transmission is affected by attacks and the network is stable, which can effectively detect multiple types of multi-stage attacks such as DDoS attacks, man-in-the-middle attacks and malware propagation. The data transmission delay is stable below 0.085 s, which verifies that the proposed algorithm can effectively improve the operational security and speed of cellular iot.
ZhangZi-yang, ZhaoJun-hui, MaXiao-ting. Large-scale device access algorithm for 5G cellular internet of things[J]. Journal of Applied Sciences, 2023, 41(4):626-635.
YinHao-ran, MiaoShi-hong, HanJi, et al. Anomaly identification method for distribution internet of things based on three-dimensional convolutional neural network[J]. Automation of Electric Power Systems, 2022, 46(1): 42-50.
LiuXiang-jun, JiangLing-yun. Abnormal traffic detection algorithm for IoT devices based on ensemble learning[J]. Application Research of Computers,2022, 39(6): 1785-1789, 1804.
DingQing-feng, LiJin-guo. A distributed abnormal traffic detection scheme in internet of things environment[J]. Computer Engineering, 2022, 48(8): 152-159.
HeQing-lin, WangLi-hong, ChenYan-jiao, et al. An automatic and real-time detection method of IoT in-the-wild vulnerability attack[J]. Journal of Beijing University of Aeronautics and Astronautics, 2024, 50(7): 2195-2205.
LuQin-zheng, ZhuXiao-juan. Research on internet of things traffic prediction based on parallel multiscale convolutional memory residual network[J]. Journal of Langfang Teachers University(Natural Science Edition), 2024, 24(1): 33-41.
ZengFan-feng, TianYu-si, WangJing-zhong. Research on early detection method of node capture attack in internet of things[J]. Computer Simulation,2022, 39(8): 477-481.
ZhangYue, TangLun, WangKai, et al. Anomaly detection algorithm of IoT devices based on GB-AEnet-FL network[J]. Application Research of Computers, 2022, 39(11): 3410-3416.
YuanDan-fei, ChenCi-fa, DongFang-min. Research on residual network of image recognition based on multiscale split[J]. Computer Engineering, 2022, 48(5): 258-262, 271.
TangYun, Peng-feiShuai, JiangPei-fan, et al. Homography estimation method based on multi-scale residual network[J]. Application Research of Computers, 2022, 39(10): 3179-3185.
ZhangYa-wen, WangZhi-hai, LiuHai-yang, et al. Time series classification algorithm based on multiscale residual full convolutional neural network[J]. Journal of Software, 2022, 33(2): 555-570.