Research on Communication Signal Modulation Recognition Based on a CCLDNN

Zijin He,Xiaodong Zeng
DOI: https://doi.org/10.3390/electronics13091604
IF: 2.9
2024-04-24
Electronics
Abstract:In this paper, a new automatic modulation recognition (AMR) method named CCLDNN (complex-valued convolution long short-term memory deep neural network) is proposed. It is designed to significantly improve the recognition accuracy of modulation modes in low signal-to-noise ratio (SNR) environments. The model integrates the advantages of existing mainstream neural networks. The phase and amplitude information of complex signals is effectively captured through a complex module in the input layer. The Squeeze-and-Excitation (SE) attention mechanism, Bi-LSTM layer, and deep convolutional layer are introduced in the feature extraction layer to gradually enhance feature expression. Among these, the introduction of LSTM enables the model to capture the sequence dependence of signals, and the application of the SE attention mechanism further improves the model's ability to focus on key features. Tests using the RadioML2016.10a dataset show that the model performs well at multiple SNR levels, achieving an average recognition accuracy of more than 80% over an SNR range of 0 dB to 18 dB. However, under the condition of a low SNR from −20 dB to −2 dB, the model still maintains a high recognition ability. The advanced CCLDNN method shows great deep learning potential in solving practical communication problems.
engineering, electrical & electronic,computer science, information systems,physics, applied
What problem does this paper attempt to address?
The problem that this paper attempts to solve is: **How to improve the accuracy of communication signal modulation recognition in a low signal - to - noise ratio (SNR) environment**. Specifically, the paper proposes a new automatic modulation recognition (AMR) method, named CCLDNN (Complex Convolutional Long - Short - Term Memory Deep Neural Network), aiming to significantly improve the modulation pattern recognition accuracy in a low SNR environment. The paper solves the problem in the following ways: 1. **Introduction of the complex module**: Effectively capture the phase and amplitude information of complex signals through the complex module. 2. **Design of the feature extraction layer**: Introduce the Squeeze - and - Excitation (SE) attention mechanism, bidirectional LSTM layer and deep convolutional layer to gradually enhance the feature expression ability. - LSTM is used to capture the time - series dependence of signals. - The SE attention mechanism further improves the model's ability to focus on key features. 3. **Optimization strategy**: Use the improved complex Adam optimizer, combined with the adaptive learning rate adjustment and weight decay strategy, to improve the model training efficiency and final performance. 4. **Experimental verification**: Through testing on the RadioML2016.10a dataset, it is proved that the model performs well at multiple SNR levels. The average recognition accuracy exceeds 80% in the SNR range from 0 dB to 18 dB, and still maintains a high recognition ability under low SNR conditions (−20 dB to −2 dB). In summary, the core objective of the paper is to solve the problem of insufficient recognition accuracy of traditional methods in a low SNR environment by designing the CCLDNN model, thereby providing a more efficient solution for actual communication problems.