Weak Signal Detection Based on WT-LSTM

In wireless communication, when the power of received signal is much weaker than the power of noise, signal detection will be very difficult. For the problem of weak signal detection, a new method is proposed in this paper, which combines the two nonlinear methods of Wavelet Transform (WT) and Long Short-Term Memory (LSTM) neural networks together, denoted as WT-LSTM. WT-LSTM uses wavelet decomposition to extract signal features and LSTM to detect weak signals, which not only makes full use of the advantages of LSTM in the classification of time series data, but also improves the information redundancy problem of detecting original signal directly using LSTM, and thereby improves the detection accuracy. Experimental results show that WT-LSTM has better detection performance compared to the existing methods.