Online IoT device identification method based on SOINN

To solve the problems of difficult dynamic model updating, insufficient training data and high computing cost in traditional device identification methods, an online Internet of Things (IoT) device identification method based on self-organizing incremental neural network (SOINN) is proposed. First, the device brand features based on document object model (DOM) tree of interactive pages are designed, which have the advantages of easy extraction, small dimensions and wide coverage of devices. The incremental supervised learning method SOINN-SVM is constructed by combining SOINN and support vector machine (SVM) to realize incremental learning of the device brand classifier, which avoids repeatedly using complete data sets to train classification models and speeds up the model updating. Then term frequency-inverse document frequency (TF-IDF) technique is used to optimize the regular matching results, which gives the model field a high weight value. The matching degree between the device to be identified and the model feature library is calculated by combining Jaro distance and weight value to realize the device model recognition. Experimental results show that the accuary of the method is 95.9%, and 37.05% more devices can be identified than other methods. The proposed method can dynamically update classification models and reduce the costs of computation and storage. © 2021, Editorial Department of Journal of Southeast University. All right reserved.