Individual identification of communication radiation sources based on Inception and LSTM network

Abstract

The difference in fingerprint characteristics of communication radiation sources is small, and it is difficult to extract characteristics for recognition using traditional machine learning algorithms, so deep learning methods are considered. LSTM is an improved recurrent neural network that is good at processing long-term sequence data. The Inception module can obtain features of different scales on the same layer. This article combines the inception structure and the LSTM network to identify 5 USRPs. The data set used in the experiment was collected by USRP and Lab VIEW. Two sets of data were collected based on the obstacles between the sending end and the receiving end, which are closer to the real environment than the data set that only uses software simulation. Compared with other network structures, this method has a higher recognition rate.