A Terminal Classification Scheme with Imbalanced Dataset Based on Low-Complexity Time-LSTM

With the rapid development of wireless communication, various kinds of mobile terminals may communicate with each other. To provide sufficient security and privacy, classification of the source terminals is generally critical. In this paper, we set up several scattered sensors collecting the received field strength and capture time for classification. To deal with the unknown propagation environment, imbalanced dataset and irregular sampling time, we propose a terminal classification scheme based on Long Short-Term Memory (LSTM) network. First, the problem of imbalanced dataset is solved by designing a preprocessing method called random interval sampling method, where the samples for class with less terminals are resampled. Then, the information of the irregular sampling time is incorporated into the classification to obtain extra benefit. Experimental results based on real-world data demonstrate that when compared with the exsiting LSTM schemes, the proposed classification model effectively utilizes the irregular time intervals and achieves excellent classification perfomance with imbalanced dataset.