A Hybrid LSTM-ResNet Deep Neural Network for Noise Reduction and Classification of V-Band Receiver Signals

• Published in: IEEE access 2022-01, Vol.10, p.1-1
• Main Authors: Arab, Homa, Ghaffari, Iman, Evina, Romaric M., Tatu, Serioja O., Dufour, Steven
• Format: Article
• Language: English
• Subjects: Algorithms; Artificial intelligence; Artificial neural networks; Bit error rate; Classification; Coders; convolutional neural network (CNN); Data transfer (computers); deep learning; Demodulation; denoising; Doppler frequency; long short-term memory (LSTM); machine learning; millimeter wave; Millimeter wave technology; Millimeter waves; modulator; Neural networks; Noise control; Noise levels; Noise reduction; Phase shift keying; Predictive models; Quadrature amplitude modulation; receiver; Receivers; recurrent neural network (RNN); Recurrent neural networks; ResNet; Sequences; signal classification; Signal processing; Signal to noise ratio; Solid modeling; White noise; Wireless communication systems; Wireless networks
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2022.3147980

Noise reduction is one of the most important process used for signal processing in communication systems. The signal-to-noise ratio (SNR) is a key parameter for minimizing the bit error rate (BER). The inherent noise in millimeter-wave systems is mainly a combination of white and phase noise. Increasing the SNR can lead to reliability and performance improvements in wireless data transfer systems. To address this issue, we propose to use a recurrent neural network (RNN) with a long short-term memory (LSTM) autoencoder architecture to achieve signal noise reduction. This design is based on a composite LSTM autoencoder with a single encoder layer and two decoder layers. A V-band receiver test bench is designed and fabricated to provide a high-speed wireless communication system. Constellation diagrams display the output signals measured for various random sequences of PSK and QAM modulated signals. The LSTM autoencoder is trained in real time using various noisy signals. The trained system is then used to reduce noise levels in the tested signals. The SNR of the designed receiver is of the order of 11.8 dB and it increases to 13.66 dB using the three-level LSTM autoencoder. Consequently, the proposed algorithm reduces the bit error rate from 10-8 to 10-11. The performance of the proposed algorithm is comparable to other noise reduction strategies. Augmented denoised signals are fed into a ResNet-152 deep convolutional network to perform the final classification. The demodulation types are classified with an accuracy of 99.93%. This is confirmed by experimental measurements.