[1712.00443] Deep Neural Network Architectures for Modulation Classification
The residual and densely connected networks (ResNet and DenseNet) also perform well although the best accuracy is limited by the depth of network, but they suggest that changing connections between layers and specially creating shortcuts between non-consecutive layers may produce better classification accuracy.
Abstract: In this work, we investigate the value of employing deep learning for the
task of wireless signal modulation recognition. Recently in , a framework
has been introduced by generating a dataset using GNU radio that mimics the
imperfections in a real wireless channel, and uses 10 different modulation
types. Further, a convolutional neural network (CNN) architecture was developed
and shown to deliver performance that exceeds that of expert-based approaches.
Here, we follow the framework of  and find deep neural network architectures
that deliver higher accuracy than the state of the art. We tested the
architecture of  and found it to achieve an accuracy of approximately 75% of
correctly recognizing the modulation type. We first tune the CNN architecture
of  and find a design with four convolutional layers and two dense layers
that gives an accuracy of approximately 83.8% at high SNR. We then develop
architectures based on the recently introduced ideas of Residual Networks
(ResNet ) and Densely Connected Networks (DenseNet ) to achieve high SNR
accuracies of approximately 83.5% and 86.6%, respectively. Finally, we
introduce a Convolutional Long Short-term Deep Neural Network (CLDNN ) to
achieve an accuracy of approximately 88.5% at high SNR.
‹Fig. 1: A building block of ResNet. (Introduction)Fig. 2: Architecture of seven-layer CNN (Simulation Setup)Fig. 3: Architecture of seven-layer ResNet. (Simulation Setup)Fig. 4: Architecture of seven-layer DenseNet. (Simulation Setup)Fig. 5: Architecture of eight-layer CLDNN. (Simulation Setup)Fig. 6: Varying hyper-parameters in CNN. Accuracies at lower SNR are similar, the four-convolutional-layer architecture delivers an accuracy of 83.8% at high SNR. (Convolutional Neural Network)Fig. 7: Best Performance at high SNR is achieved with a fourconvolutional-layer DenseNet. (Densely Connected Network)Fig. 8: Classification performance comparison between candidate architectures. CLDNN and DenseNet outperform other models with best accuracies of 88.5% and 86.6%, respectively. (CLDNN)Fig. 9: The confusion matrix of CLDNN at SNR=18dB. (CLDNN)Fig. 10: Validation loss descents quickly in all three models, but losses of DenseNet and ResNet reach plateau earlier than that of CNN. (Discussion)›
[1703.09197] Deep Architectures for Modulation Recognition[1901.05850] Fast Deep Learning for Automatic Modulation Classification[1803.03759] Speech Recognition: Keyword Spotting Through Image Recognition[1909.03050] Sequential Convolutional Recurrent Neural Networks for Fast Automatic Modulation Classification[1909.03050v1] Sequential Convolutional Recurrent Neural Networks for Fast Automatic Modulation Classification
[1608.06993] Densely Connected Convolutional Networks