SMOTE-based Category Imbalance for Radar Radiation Source Sorting and Identification

Abstract

The aim of this study is to realize the problem of sorting and identifying radar radiation sources with category imbalance. In this paper, a method based on the Synthetic Minority Oversampling Technique (SMOTE) is presented. In the study, the SMOTE technique is applied to generate a larger number of minority samples in the data sets. In the study, the SMOTE technique is applied to generate a larger number of minority samples in the database In the study, the SMOTE technique is applied to generate a larger number of minority samples in the data sets, so that the number of majority and minority categories are gradually balanced, and then the parameters of the radar radiation source signal (direction of arrival (DOA), pulse width (PW), pulse repetition frequency (PRF), and radar frequency (RF)) are selected as characteristics for classification, using decision trees, random forest, and XGboost three supervised learning algorithms are used to classify the rebalanced data sets. The experimental results show that the STOME algorithm excels in category-imbalanced radar source sorting identification, improving the problem of low accuracy caused by the presence of a few class samples in traditional methods.