A Specific Emitter Identification System Design for Crossing Signal Modes in the Air Traffic Control Radar Beacon System and Wireless Devices

Abstract

To improve communication stability, more wireless devices transmit multi-modal signals while operating. The term 'modal' refers to signal waveforms or signal types. This poses challenges to traditional specific emitter identification (SEI) systems, e.g., unknown modal signals require extra open-set mode identification; different modes require different radio frequency fingerprint (RFF) extractors and SEI classifiers; and it is hard to collect and label all signals. To address these issues, we propose an enhanced SEI system consisting of a universal RFF extractor, denoted as multiple synchrosqueezed wavelet transformation of energy unified (MSWTEu), and a new generative adversarial network for feature transferring (FTGAN). MSWTEu extracts uniform RFF features for different modal signals, FTGAN transfers different modal features to a recognized distribution in an unsupervised manner, and a novel training strategy is proposed to achieve emitter identification across multi-modal signals using a single clustering method. To evaluate the system, we built a hybrid dataset, which consists of multi-modal signals transmitted by various emitters, and built a complete civil air traffic control radar beacon system (ATCRBS) dataset for airplanes. The experiments show that our enhanced SEI system can resolve the SEI problems associated with crossing signal modes. It directly achieves 86% accuracy in cross-modal emitter identification using an unsupervised classifier, and simultaneously obtains 99% accuracy in open-set recognition of signal mode.

Keywords: air traffic control radar beacon system (ATCRBS); multiple modal; radio frequency fingerprint (RFF); specific emitter identification (SEI); unify features.

Figure 1

The process of a traditional SEI system for multi-modal signals and our SEI system for multi-modal signals. Our system aims to unify the extractor and classifier, which improves the efficiency of the system. With this improvement, the system decreases labeling work.

Figure 2

The MSWTEu process. The first column waveform is the time-domain wave; in the second to third columns, the blue wave represents $SW{T}_{ig}$, and the red wave represents $SW{T}_{ih}$.

Figure 3

The emitter label identification procedure for a single mode (Mode 1) and four different modes (Mode 1/2/3/4). The top right part is the SEI classifier trained for Mode 1 under supervision, which outputs the emitter ID. The bottom right is our multi-modal SEI system, which recognizes the emitter ID for all modal signals. Our multi-modal SEI system uses FTGAN to transfer different mode features to Mode 1 distribution and shares parameters with the Mode 1 SEI classifier.

Figure 4

The architectural comparison between normal GAN and FTGAN.

Figure 5

The waveform feature comparison between MSWT and MSWTEu; (a,c,e,g) are the MSWT waveform features of rm1, wifi, ads-b, and rm2; (b,d,f,h) are the MSWTEu waveform features of rm1, wifi, ads-b, and rm2; the red boxes are noises not decreased by the MSWT, which do not exist in the MSWTEu waveform features; the horizontal axis displays the frequency and the vertical axis displays the amplitude.

Figure 6

The SEI accuracy comparison between MSWTEu, MSWTE, and ISWTE in identifying emitters by ads-b signals under different SNRs.

Figure 7

Boxplot representation of MSWT (a) and MSWTEu (b) for 12 devices; the values of the same red boxes show a similarity which crosses different modes in (a) but not crosses different modes in (b); the horizontal axis represents the signal mode while the vertical axis indicates the value of each feature. For visualization purposes, these features have been reduced to 1 dimension using T-distributed stochastic neighbor embedding (TSNE).

Figure 8

The accuracy comparison for various methods in recognizing signal modes.

Figure 9

The accuracy confusion matrix for various methods in recognizing signal modes, (a) is MAp, (b) is MSWT, (c) is MSWTEu, and the darker color represents the higher accuracy.

Figure 10

MSWT and MSWTEu performances in recognizing 3 emitters for the same mode, (a) and MSWTEu and MSWTEu+FTGAN performances in recognizing 12 emitters for multiple modes (b).

Figure 11

The SEI accuracy through multi-modal signals based on different methods.

Figure 12

The comprehensive SEI accuracy confusion matrix through multi-modal signals based on MSWTEu+FTGAN, and the darker color represents the higher accuracy.