Abstract
Recently, studies to utilize artificial neural networks in radar detection have been actively conducted. In this paper, we compare the performances of SOTA (State Of The Art) neural networks when applying to radar detection. The detection networks consist of two paths to detect the range and Doppler frequency. Among the networks used, HRNet shows the best performance for the range detection, and EfficientNet shows the best performance for Doppler frequency prediction. Although the detection probability is slightly lower than the conventional radar signal processing result, it is somewhat insensitive to the detection threshold, and the false alarm rate is quite low. Therefore, it is expected that the SOTA networks can be well utilized in radar detection.
| Translated title of the contribution | Analysis of Radar Signal Detection Performance Using Deep Convolutional Neural Network |
|---|---|
| Original language | Korean |
| Pages (from-to) | 439-447 |
| Number of pages | 9 |
| Journal | 방송공학회 논문지 |
| Volume | 28 |
| Issue number | 4 |
| DOIs | |
| State | Published - 2023 |