Application of novel architectures for Modulation Recognition
Yujie Sang, Li Li · 2018
In this paper, we investigate the latest deep learning framework for wireless modulation recognition. The deep learning method can substantially improve the recognition performance with relatively small scale network. We have incorporated the modified convolutional neural network (CNN) and Long Short-term Memory (LSTM) to further improve the performance and reduce the complexity of the deep learning framework with lower recognition error rate. Results show that the proposed simple LSTM network can achieve better performance than other network structures. Although the capsule network is inferior to other network structures, it may have potential ability to solve this problem.