Deep Anomaly Detection with Attention (DADA): A Novel Approach for Identifying Multipath Interference in Radar Signals

Table 1

Performance metrics for the four methods (test set with single multipath signal).

Method	LFM (L)			NLFM (N)			BPSK (B)			QPSK (Q)
	P	R	F1	P	R	F1	P	R	F1	P	R	F1
CCF	0.709	1.000	0.829	0.687	1.000	0.814	0.665	0.995	0.797	0.689	1.000	0.816
SC	0.657	1.000	0.793	0.678	1.000	0.808	0.586	1.000	0.739	0.664	1.000	0.798
WRFCCF	0.707	0.990	0.825	0.697	0.990	0.818	0.670	0.935	0.460	0.604	0.980	0.748
AnoGAN	0.673	0.970	0.795	0.803	0.960	0.874	0.498	0.995	0.664	0.593	0.870	0.705
EGBAD	0.668	0.995	0.799	0.921	0.995	0.956	0.505	0.990	0.669	0.619	0.975	0.757
GANomaly	0.872	1.000	0.931	0.854	0.899	0.876	0.576	0.953	0.719	0.898	0.605	0.723
DADA	1.000	0.980	0.990	1.000	0.960	0.979	0.731	0.925	0.817	0.866	0.905	0.885