Research Article
Cyber-Physical Security with RF Fingerprint Classification through Distance Measure Extensions of Generalized Relevance Learning Vector Quantization
Table 3
Classification and verification performance for each algorithm.
| Method | Performance results | Classification | Verification at SNR = 20 dB | GSNR (dB) at %C = 90% | RAP | %Aut | AUCM | TNG | TST | TNG | TST |
| Cosine GRLVQI (proposed herein) | +5.23 | +6.00 | 1.14 | 1.14 | 66 | 0.973 | GRLVQI-SD optimized [28] | +5.16 | +5.05 | 1.16 | 1.17 | 66 | 0.965 | GRLVQI (baseline) [29] | +3.72 | +3.32 | 1.14 | 1.13 | 33 | 0.936 | MDA [29] | +1.68 | 0.00 | 1.23 | 1.0 | 100 | 0.971 |
|
|
Performance results bolded are the best values and those within 5% of the best values.
|