Design of Automatic Modulation Classification Based on Deep Learning Technique Applying Extended Frame 


Vol. 46,  No. 8, pp. 1227-1236, Aug.  2021
10.7840/kics.2021.46.8.1227


PDF
  Abstract

In this paper, a new deep learning-based automatic modulation classification technique is proposed for automatic modulation classification that is applied to cognitive radio networks. The proposed model is designed by Convolutional Neural Network (CNN) technique, which contains skip connection structure of ResNet to minimize the vanishing gradient problem and convolutional layers with asymmetric kernels to reduce computation complexity. In general, complex signals are divided into 2 × N to utilize as input size of the CNN. However, in this paper, by extending the kept frame from a transmitter 4 × N size is adopted for the proposed CNN. For the frame extension method there are three steps. First, the present frame is copied. Second, the order horizontally is reversed. Finally, it is connected to the existing frame. To evaluate the performance of the proposed CNN model the DeepSig:RadioML 2018.01A dataset with 24 modulation schemes was used, and the accuracy performance and the computational complexity are compared to the latest deep learning models ResNet, MCNet and LCNN. Through the simulation results, the proposed model has better prediction accuracy performance in the entire signal-to-noise ratio (SNR) region than the conventional model, where it showed at least 2.5% higher accuracy at SNR 10 dB.

  Statistics
Cumulative Counts from November, 2022
Multiple requests among the same browser session are counted as one view. If you mouse over a chart, the values of data points will be shown.


  Cite this article

[IEEE Style]

S. Kim, C. Moon, J. Kim, D. Kim, "Design of Automatic Modulation Classification Based on Deep Learning Technique Applying Extended Frame," The Journal of Korean Institute of Communications and Information Sciences, vol. 46, no. 8, pp. 1227-1236, 2021. DOI: 10.7840/kics.2021.46.8.1227.

[ACM Style]

Seung-Hwan Kim, Chang-Bae Moon, Jae-Woo Kim, and Dong-Seong Kim. 2021. Design of Automatic Modulation Classification Based on Deep Learning Technique Applying Extended Frame. The Journal of Korean Institute of Communications and Information Sciences, 46, 8, (2021), 1227-1236. DOI: 10.7840/kics.2021.46.8.1227.

[KICS Style]

Seung-Hwan Kim, Chang-Bae Moon, Jae-Woo Kim, Dong-Seong Kim, "Design of Automatic Modulation Classification Based on Deep Learning Technique Applying Extended Frame," The Journal of Korean Institute of Communications and Information Sciences, vol. 46, no. 8, pp. 1227-1236, 8. 2021. (https://doi.org/10.7840/kics.2021.46.8.1227)

Search
(IN TITLE, AUTHOR, ABSTRACT,KEYWORDS)

Advanced Search ▼

POPULAR KEYWORDS
(TOP 10 KEYWORDS)

Recent Publications
(LAST 3 YEARS)






pISSN : 1226 - 4717
eISSN : 2287 - 3880