Single-Channel Blind Source Separation of Spatial Aliasing Signal Based on Stacked-LSTM

Abstract

Aiming at the problem of insufficient separation accuracy of aliased signals in space Internet satellite-ground communication scenarios, a stacked long short-term memory network (Stacked-LSTM) separation method based on deep learning is proposed. First, the coding feature representation of the mixed signal is extracted. Then, the long sequence input is divided into smaller blocks through the Stacked-LSTM network with the attention mechanism of the SE module, and the deep feature mask of the source signal is trained to obtain the Hadamard product of the mask of each source and the coding feature of the mixed signal, which is the encoding feature representation of the source signal. Finally, characteristics of the source signal is decoded by 1-D convolution to to obtain the original waveform. The negative scale-invariant source-to-noise ratio (SISNR) is used as the loss function of network training, that is, the evaluation index of single-channel blind source separation performance. The results show that in the single-channel separation of spatially aliased signals, the Stacked-LSTM method improves SISNR by 10.09∼38.17 dB compared with the two classic separation algorithms of ICA and NMF and the three deep learning separation methods of TasNet, Conv-TasNet and Wave-U-Net. The Stacked-LSTM method has better separation accuracy and noise robustness.

A Robust Hybrid Neural Network Architecture for Blind Source Separation of Speech Signals Exploiting Deep Learning

“…Researchers have made significant contributions in developing a wide range of systems and algorithms aimed at signal separation. However, despite these advancements, there are persistent challenges that impede the achievement of precise and timely separation of all signals [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36]. Extensive research efforts have been devoted to the field of BSS, leading to the proposal of numerous techniques that utilize a range of existing methods [37], [38], [39], [40], [41].…”

A Dual Attention-Based Autoencoder Model for Fetal ECG Extraction From Abdominal Signals

“…The average root mean square error (RMSE) obtained using this proposed method was reported to be less than 0.037. Zhao et al [15] used a stacked long short-term memory network (Stacked-LSTM) to improve blind source separation performance. Their proposed model used an auto-encoder with a separation part which includes the stacked-LSTM with the attention mechanism of the squeeze-and-excitation (SE) module.…”

“…An automatic windowing procedure is also performed to generate the labelled data for training. According to the literature, [14], [15], the attention modules can help deep learning models to find and extract specific signal patterns. Therefore, the proposed auto-encoder model in this study is equipped with novel attention modules to focus on the extraction of FECG patterns.…”

Improved Swarm Intelligent Blind Source Separation Based on Signal Cross-Correlation

“…The separated signal is . The mathematical model of blind source separation [ 1 , 2 , 3 , 5 , 16 , 17 , 19 , 20 , 21 , 23 , 41 , 42 , 43 , 44 , 45 , 46 ] is a linear mixture model, as shown in Equation (1): where A is the linear mixing matrix of , and n represents the number of source signals. According to the de-correlation and non-Gaussian criteria, the separation matrix W is solved, and then the separated signal is extracted from the , which can be expressed as Equation (2): where is the separation matrix of .…”