Visual analysis method for RF fingerprint based on important region localization and masking

Wenbin Liu; Pingzhi Fan; Jiahuang Yang; Yukai Li; Yuhao Wang; Hua Meng

doi:10.11884/HPLPB202436.230380

Journals >High Power Laser and Particle Beams >Volume 36 >Issue 4 >Page 043019 > Article

High Power Laser and Particle Beams
Vol. 36, Issue 4, 043019 (2024)

Visual analysis method for RF fingerprint based on important region localization and masking

Wenbin Liu^1、2, Pingzhi Fan¹, Jiahuang Yang³, Yukai Li³, Yuhao Wang³, and Hua Meng^3、*

Author Affiliations

¹School of Information Science & Technology, Southwest Jiaotong University, Chengdu 611756, China

²The 30th Research Institute of China Electronics Technology Group Corporation, Chengdu 610041, China

³School of Mathematics, Southwest Jiaotong University, Chengdu 611756, China

show less

DOI: 10.11884/HPLPB202436.230380 Cite this Article

Wenbin Liu, Pingzhi Fan, Jiahuang Yang, Yukai Li, Yuhao Wang, Hua Meng. Visual analysis method for RF fingerprint based on important region localization and masking[J]. High Power Laser and Particle Beams, 2024, 36(4): 043019 Copy Citation Text

show less

Fig. 1. Network architecture of MiddleResNet and LittleResNet

Download full size | View in the Article

Fig. 2. Comparison of time-domain pulse signal heat maps between MiddleResNet and LittleResNet

Download full size | View in the Article

Fig. 3. Important areas of emitters when using MiddleResNe and LittleResNet

Download full size | View in the Article

Fig. 4. Test results by continuous mask and different samples comparison

Download full size | View in the Article

model	accuracy				parameter number/10⁶
model	5 dB	9 dB	14 dB	all data	parameter number/10⁶
LittleResNet	0.7605	0.8775	0.9605	0.8812	0.153
MiddleResNet	0.7665	0.8990	0.9685	0.8930	0.254
Transformer	0.8670	0.9493	0.995	0.9439	1.021

Table 1. Performance comparison results of three models

emitter	Transformer	MiddleResNet	LittleResNet
1#	1(sample23)	4(sample3,5,12,17)	4(sample5,10, 12, 23)
2#	0	7(sample3,5,7,14,19,22,28)	6(sample3,6,7,19,22,28)
3#	0	0	0
4#	1(sample19)	2(sample7,19)	2(sample7,19)
5#	1(sample9)	1(sample14)	0
6#	0	0	0
7#	0	0	0
8#	0	0	0
9#	0	0	0
10#	0	0	0

Table 2. Number and sequence of mispredicted samples

sample	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30
no masking	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3
1^st masking	3	3	2	3	3	3	3	3	1	3	3	1	3	3	3	3	3	3	3	3	3	3	3	3	2	3	2	3	3	3
2^nd masking	3	3	2	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	2	3	3	3
3^rd masking	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	3	2	3	3	3	2	3	3	3

Table 3. Prediction results when masking 3# and using the MiddleResNet model

emitter	MiddleResNet					LittleResNet
emitter	no masking	1^stmasking	2^ndmasking	3^rdmasking	emitter(s) mainly mispredicted to	no masking	1^stmasking	2^ndmasking	3^rdmasking	emitter(s) mainly mispredicted to
1#	4	27	29	29	2#	4	10	10	10	2#
2#	7	16	19	19	1#、3#	6	10	14	10	1#、3#
3#	0	5	2	2	2#、1#	0	0	0	0	——
4#	2	13	27	29	3#	2	12	16	10	3#、2#、5#、1#
5#	1	20	29	29	3#、4#	0	5	14	17	3#
6#	0	13	28	30	7#、10#	0	15	23	25	7#、10#
7#	0	7	9	16	10#、3#	0	1	1	7	10#、3#
8#	0	9	20	28	10#、7#、9#、3#	0	7	13	16	10#、7#、6#、9#
9#	0	8	14	17	10#、7#、3#	0	6	13	16	10#
10#	0	2	11	15	7#、3#	0	2	3	4	7#

Table 4. Number of mispredicted samples and emitter(s) mainly mispredicted to when masking different times

Download Citation

Save the article for my favorites

Paper Information