In this paper, we designed a ultra lite CNN (ULCNN) (9,751 trainable parameters and 0.2M MACCs) for AMC, and its simulation is based on RML2016.10a
http://arxiv.org/abs/2208.04659
keras=2.1.4 tensorflow=1.14
MCLDNN [1]
SCNN [2]
MCNet [3]
PET-CGDNN [4]
ULCNN is the proposed structure.
The model weights are given in "model/"
RML2016.10a
Train/val/test samples: 77000/33000/110000
https://pan.baidu.com/s/1T36jgWlZ3oWmFWYpQLyiZg, passwd：f7qy or run dataset2016.py
[1] J. Xu, C. Luo, G. Parr and Y. Luo, "A Spatiotemporal Multi-Channel Learning Framework for Automatic Modulation Recognition," in IEEE Wireless Communications Letters, vol. 9, no. 10, pp. 1629-1632, Oct. 2020, doi: 10.1109/LWC.2020.2999453.
[2] X. Fu et al., "Lightweight Automatic Modulation Classification Based on Decentralized Learning," in IEEE Transactions on Cognitive Communications and Networking, vol. 8, no. 1, pp. 57-70, March 2022, doi: 10.1109/TCCN.2021.3089178.
[3] T. Huynh-The, C. Hua, Q. Pham and D. Kim, "MCNet: An Efficient CNN Architecture for Robust Automatic Modulation Classification," in IEEE Communications Letters, vol. 24, no. 4, pp. 811-815, April 2020, doi: 10.1109/LCOMM.2020.2968030.
[4] F. Zhang, C. Luo, J. Xu and Y. Luo, "An Efficient Deep Learning Model for Automatic Modulation Recognition Based on Parameter Estimation and Transformation," in IEEE Communications Letters, vol. 25, no. 10, pp. 3287-3290, Oct. 2021, doi: 10.1109/LCOMM.2021.3102656.
Note that our code is partly based on leena201818, wzjialang, ThienHuynhThe and Richardzhangxx.
Thanks for your great works!