IOT device identification solution based on unsupervised to semi-supervised
Rongxin Liu, Qin Li, Weiyuan Li, Zhiqiang Li, Tao Sun, Lu Lu · 2021 IEEE International Conference on Data Science and Computer Application (ICDSCA) · 2021
With the development of 5G, the types and numbers of IoT terminals will become more abundant. Different types of IoT terminals have large differences in behavior and network requirements. The refined operation and maintenance based on IoT application scenarios is an important for operators. The identification of the IoT terminal type is the first step to achieve refined operation and maintenance. In order to address difficulties of identifying IOT terminal type and lack of labeled data, this paper proposes a solution to fulfill the IoT terminal type based on network data using AI algorithm. This solution is to extract the behavioral patterns of IoT terminals based on network data, use a combination of semi-supervised and unsupervised methods to fulfill the labeling of training data, and use AI algorithms to identify IoT terminal type. The verification results show that the accuracy in identifying bus IoT from the APN of the IoT terminal hybrid access reaches 97.87%, which proved the effectiveness of the solution.