Specific Emitter Identification Based on a Hybrid Deep Neural Network for ACARS Authentication

Table 3

Identification accuracy and computational complexity concerning various hyperparameter settings for local feature learning.

Number of RSUs	Identification accuracy	Computational complexity (MFLOPs)
4	0.937	0.83
8	0.968	1.66
12	0.980	2.50
16	0.982	3.33
20	0.982	4.16