Research Article
Automatic Modulation Classification Based on Deep Learning for Software-Defined Radio
Table 1
Test accuracy of baseline and the proposed method.
| SNR (dB) | CNN (%) | ResNet (%) | Inception (%) | LSTM (%) | MentorNet (%) | Proposed method (%) |
| ā10 | 25.71 | 26.92 | 23.61 | 29.56 | 24.86 | 30.40 | ā4 | 54.32 | 48.53 | 50.86 | 58.46 | 62.45 | 74.30 | 0 | 77.64 | 75.41 | 73.65 | 83.19 | 76.36 | 90.81 | 4 | 80.53 | 80.25 | 78.87 | 85.96 | 78.59 | 92.84 | 10 | 81.61 | 81.69 | 79.91 | 87.09 | 80.16 | 93.29 |
|
|