綜觀軍事通訊發展，在軍事電子戰應用上，針對戰場上的頻譜監控與訊號情報蒐集，如何在高複雜電磁環境下截獲敵方未知通聯訊號，快速的完成偵測、辨識、解譯，以即時獲取敵方情資，在電子戰中至關重要，其中訊號調變類型自動分類，為軍事通訊截收中的關鍵技術。
　　傳統基於特徵擷取(feature extraction)的訊號調變分類的方法，需要事先分析出各種特徵參數，再利用決策樹(decision tree)或各種機器學習(machine learning)的方式，從擷取到的特徵資料中訓練出有效的分類模型，此種方式仰賴人為擷取的特徵能夠確實提供訊號分類所需的完整資訊，然而面對通道的各種變化，人工分析得到的特徵值(expert feature)往往會受到干擾而造成分類效果不佳。
　　本文提出基於卷積網路(Convolutional Neural Network)之調變分類技術，神經網路可從訓練資料(training data)中自我學習(learning from data)，自動擷取特徵並分類，實驗結果顯示，深度卷積神經網路的分類方式有更好的抗干擾性，我們綜合了各個測試的成果，提出的模型在SNR為0dB~20dB 的範圍內，調變分類預測的準確度達到94.05%的不錯表現。

While investigating the development of military communication in the application of military electronic warfare, how to detect, identify and decode signals of interesting in the high-complex electromagnetic environment is extremely important. Automatic modulation recognition and classification has become a necessary technology in military electronic warfare.

Based on feature extraction, traditional modulation classification require prior analysis of various feature parameters, and then use decision trees or machine learning methods to extract features. The classification model is trained based on the captured features. This method relies on the expert features
providing sufficient information for signal classification. However, in the face of varied communication channel, the artificial expert features often be interfered and causes poor classification results.

This paper proposes a modulation classification technique based on Convolutional Neural Network. The neural network can learn from training data, extract features and classify signals automatically. The experimental results show that modulation classification using convolutional neural network provide better anti-interference performance. Analyses show that the proposed model yields an average classification accuracy of 94.05% at varying SNR conditions ranging from 0dB to 20dB.

[1]A. K. Nandi and E. E. Azzouz, “Algorithms for automatic modulation recognition of communication signals,” IEEE Transactions on Communications, vol. 46, no. 4, pp. 431-436, Apr 1998.
[2]林聰岷，《使用高階統計法則實現相位鍵移調變訊號分類作業》，碩士論文，國立臺灣大學電信工程學研究所，2012。
[3]C. M. Spooner, “On the utility of sixth-order cyclic cumulants for RF signal classification,” Conference Record of Thirty-Fifth Asilomar Conference on Signals, Systems and Computers, vol. 1, pp. 890-897, 2001.
[4]R. M. Al-Makhlasawy, M.M.A. Elnaby, H. A. Al-Khobby, E. M. El-Rabaie and F. E. A. El-samie, “Automatic modulation recognition in OFDM systems using cepstral analysis and support vector machines,” J. Telecomm. Syst. Manag., vol. 1, no. 3, pp.1-7, 2012
[5]K. Triantafyllakis, M. Surligas, G. Vardakis and S. Papadakis, “Phasma: An automatic modulation classification system based on Random Forest,” 2017 IEEE International Symposium on Dynamic Spectrum Access Networks (DySPAN), pp. 1-3, 2017
[6]J.C. Lin and H.Y. Hsu, “Timing-delay and frequency-offset estimations for initial synchronisation on time-varying Rayleigh fading channels,” IET Commun., vol. 7, iss. 6, pp. 562-576, 2013.
[7]J.C. Lin, “A frequency offset estimation technique based on frequency error characterization for OFDM communication on time-varying multipath fading channels,” IEEE Trans. Vehic. Technol., vol. 56, no. 3, pp. 1209-1222, May 2007.
[8]K.P. Chou, J.-C. Lin and H. V. Poor, “Disintegrated channel estimation in filter-and-forward relay networks,” IEEE Trans. Commun., vol. 64, no. 7, pp. 2835-2847, Jul. 2016.
[9]J.C. Lin, “Least-squares channel estimation for mobile OFDM communication on time-varying frequency-selective fading channels,” IEEE Trans. Vehic. Technol., vol. 57, no. 6, pp. 3538-3550, Nov. 2008.
[10]J.C. Lin, “Least-squares channel estimation assisted by self-interference cancellation for mobile PRP-OFDM applications,” IET Commun., vol. 3, iss. 12, pp.1907-1918, Dec. 2009.
[11]T. J. O’Shea, J. Corgan and T. C. Clancy, “Convolutional radio modulation recognition networks,” International Conference on Engineering Applications of Neural Networks, vol. 629, pp. 213-226, 2016
[12]W. Yongshi, G. Jie, L. Hao, L. Li, W. Zhigang and W. Houjun, “CNN-based modulation classification in the complicated communication channel,” 2017 13th IEEE International Conference on Electronic Measurement & Instruments (ICEMI), pp. 512-516, 2017
[13]J. Zhang, Y. Li and J. Yin, “Modulation classification method for frequency modulation signals based on the time–frequency distribution and CNN,” IET Radar, Sonar & Navigation, vol.12, pp. 244-249, 2017
[14]M. Kulin, T. Kazaz, I. Moerman and E. d. Poorter, “End-to-end Learning from Spectrum Data: A Deep Learning approach for Wireless Signal Identification in Spectrum Monitoring applications,” IEEE Access, vol. 6, pp.18484-48501, 2018
[15]T. J. O’Shea, N. West, M. Vondal and T. C. Clancy, "”emi-supervised radio signal identification,” 2017 19th International Conference on Advanced Communication Technology (ICACT), pp. 33-38, 2017.
[16]D. Hong, Z. Zhang and X. Xu, “Automatic modulation classification using recurrent neural networks,” 2017 3rd IEEE International Conference on Computer and Communications (ICCC), pp. 695-700, 2017
[17]S. Rajendran, W. Meert and D. Giustiniano, V. Lenders and S. Pollin, “Distributed deep learning models for wireless signal classification with low-cost spectrum sensors,” arXiv:1707.08908, 2017
[18]N.E. West and T. O'Shea, "Deep Architectures for Modulation Recognition," 2017 IEEE International Symposium on Dynamic Spectrum Access Networks (DySPAN), pp. 1-6, 2017.
[19]X. Liu, D. Yang and A. E. Gamal, “Deep Neural Network Architectures for Modulation Classification,” arXiv:1712.00443, 2017
[20]Y. S. Abu-Mostafa, M. Magdon-Ismail and H.T. Lin, (2016, Dec 26).Machine Learning Foundations. [Online]. Available:https://www.csie.ntu.edu.tw/~htlin/mooc/doc/10_present.pdf
[21]V. Nair and G.E. Hinton, "Rectified Linear Units Improve Restricted Boltzmann Machines," Proceedings of the 27th international conference on machine learning (ICML-10), pp. 807-814, 2010.
[22]Y. LeCun, C. Cortes and C.J.C. Burges. (1998).The MNIST Database. [Online]. Available: http://yann.lecun.com/exdb/mnist
[23]A Krizhevsky, V Nair and G Hinton. (2009, Nov 23).The CIFAR-10 dataset. [Online]. Available: https://www.cs.toronto.edu/~kriz/cifar.html.
[24]Y. Lecun, L. Bottou, Y. Bengio and P. Haffner, “Gradient-based learning applied to document recognition,” Proceedings of the IEEE, vol. 86, no. 11, pp. 2278-2324, 1998.
[25]H. Lee, R. Grosse, R. Ranganath, and A. Y. Ng, “Unsupervised learning of hierarchical representations with convolutional deep belief networks,” Communications of the ACM, vol. 54, no. 10, pp. 95-103, 2011.
[26]F. F. Li, J. Johnson and S. Yeung. (2017, May 4). Recurrent Neural Networks. [Online]. Available: http://cs231n.stanford.edu/slides/2017/cs231n_2017_lecture10.pdf
[27]R. Socher, A. Perelygin, J. Wu, J. Chuang, C. D. Manning, A. Ng and C Potts, “Recursive deep models for semantic compositionality over a sentiment treebank,” Proceedings of the 2013 conference on empirical methods in natural language processing, pp. 1631-1642, 2013.
[28]R. Socher, K. Clark and A. See. (2018, Feb 2). Natural Language Processing with Deep Learning. [Online]. Available:http://cse.iitkgp.ac.in/~sudeshna/courses/DL18/nlp1-9April2018.pdf
[29]D. Jurafsky and J. H. Martin. (2017, Aug 28). Speech and Language Processing. [Online]. Available: https://web.stanford.edu/~jurafsky/slp3/ed3book.pdf
[30]S. Hochreiter and J. Schmidhuber, “Long short-term memory,” Neural computation, vol. 9, no. 8, pp. 1735–1780, 1997.
[31]C. Olah. (2015, Aug 27). Understanding LSTM Networks. [Online]. Available: http://colah.github.io/posts/2015-08-Understanding-LSTMs
[32]T. O'Shea and J. Shea. (2017, Dec 13). Open Radio Machine Learning Datasets for Open Science. [Online]. Available: https://www.deepsig.io/datasets
[33]T. J. O’Shea and N. West, “Radio machine learning dataset generation with gnu radio,” Proceedings of the GNU Radio Conference, vol. 1, no. 1, 2016.
[34]K. Simonyan and A. Zisserman, "Very deep convolutional networks for large-scale image recognition," arXiv:1409.1556, 2014.
[35]N. Srivastava, G. Hinton, A. Krizhevsky, I. Sutskever and R. Salakhutdinov, "Dropout: A simple way to prevent neural networks from overfitting," The Journal of Machine Learning Research, vol. 15, no. 1, pp. 1929-1958, 2014.
[36]D. Kingma and J. Ba, “Adam: A method for stochastic optimization,” arXiv:1412.6980, 2014
[37]S. J. Pan and Q. Yang. “A survey on transfer learning,” IEEE Transactions on knowledge and data engineering, vol. 22, no. 10, pp. 1345-1359, 2010.