Enhancing Industrial Wireless Communication Security Using Deep Learning Architecture-Based Channel Frequency Response

Table 4

Result comparison under different SNRs.

Model	SNR (dB)	Accuracy	Recall	Precision	F1-score
CNN	0	0.9850	0.9843	0.9851	0.9847
	5	0.9978	0.9978	0.9979	0.9978
	−5	0.9964	0.9962	0.9965	0.9964
	10	0.9971	0.9971	0.9971	0.9971
	−10	0.9926	0.9926	0.9928	0.9927
	15	0.9981	0.9979	0.9981	0.9980
	−15	0.9922	0.9919	0.9929	0.9924
	20	0.9969	0.9969	0.9971	0.9970
	−20	0.9931	0.9921	0.9932	0.9926
LSTM	0	0.9954	0.9954	0.9954	0.9954
	5	0.9101	0.9011	0.9191	0.9100
	−5	0.8867	0.8683	0.9009	0.8843
	10	0.9258	0.9208	0.9325	0.9266
	−10	0.8786	0.8599	0.8962	0.8777
	15	0.9358	0.9315	0.9399	0.9357
	−15	0.8774	0.8628	0.8946	0.8784
	20	0.9397	0.9364	0.9436	0.9400
	−20	0.9014	0.8924	0.9129	0.9025
RBMs	0	0.8841	0.8976	0.9151	0.9063
	5	0.8354	0.7891	0.8947	0.8386
	−5	0.7904	0.7277	0.8749	0.7945
	10	0.8574	0.8280	0.9100	0.8671
	−10	0.7904	0.7195	0.8788	0.7912
	15	0.8936	0.8803	0.9301	0.9045
	−15	0.7421	0.6618	0.8445	0.7421
	20	0.9016	0.9022	0.9334	0.9175
	−20	0.7250	0.6311	0.8407	0.7210