講演抄録/キーワード |
講演名 |
2021-05-21 10:00
深層学習(ResNet)を利用したCNNの変調方式(OFDM及びシングルキャリア)識別評価 ○井手輝二(鹿児島高専)・Rozeha A Rashid・Leon Chin・M A Sarijari・Rubita Sudirman(マレーシア工科大) SR2021-9 |
抄録 |
(和) |
(まだ登録されていません) |
(英) |
In this study, we investigate and present a deep residual learning for modulation classification. The simulation results show the degradation problem that was exposed due to an increase in network depth and the saturation of accuracy in the modified conventional CNN; however, the proposed CNN has no such degradation. Therefore, the processing burden of the conventional CNN is much larger than the proposed CNN. In the simulation results, the proposed CNN framework achieves almost the same modulation classification accuracy as the normal CNN framework when reducing the processing burden in the proposed one. The better simulation results are shown by adjustment of the parameters using the proposed method in the case of OFDM and single carrier modulation types. |
キーワード |
(和) |
/ / / / / / / |
(英) |
CNN / cognitive radio / residual learning / modulation classification / / / / |
文献情報 |
信学技報, vol. 121, no. 30, SR2021-9, pp. 57-64, 2021年5月. |
資料番号 |
SR2021-9 |
発行日 |
2021-05-13 (SR) |
ISSN |
Online edition: ISSN 2432-6380 |
著作権に ついて |
技術研究報告に掲載された論文の著作権は電子情報通信学会に帰属します.(許諾番号:10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) |
PDFダウンロード |
SR2021-9 |