Individual Identification of Electronic Equipment Based on Turn-on Transient Signal

This paper presents a novel method of classifying and identifying Individual electronic devices by recognizing transient Turn-on signal based on machine learning technology, which has significant applications in electronic equipment monitoring system. In the research, four different electronic devices are selected as the radiation sources, whose Turn-on signals, 4961 samples in total, acquired and then analyzed by six classifiers. As separating 4961 transient signals from raw signals is difficult, an effective signal segment interception algorithm based on sliding window adaptive variance is proposed. Besides, we also construct a hybrid template of features, which contains 38 features in total. Then, send the template to six models for training, including Support Vector Machine (SVM), Random Forest (RF), Gradient Boosting Decision Tree (GBDT), Linear Discriminant Analysis(LDA), Decision Tree(DT) and Extreme Gradient Boosting(XGBoost). The average accuracy rate of the six models is 96.21%. Finally, a hard voting method adopted indicating that it not only increases the average accuracy rate to 97.91% but also improves the stability of the system with a few samples.