Deep Learning Detection Method for Signal Demodulation in Short Range Multi-path Channel

摘要

Signal demodulation in short range multi-path channel plays an important role in communication system. The existed wireless communication system in short range multi-channel achieve signal demodulation by using a equalizer to minimize the effect of inter-code crosstalk caused by the channel before the signal detection. However, channel equalization methods are either with high complexity or a waste of frequency resource. In this paper, we propose a deep learning based detection method for signal demodulation. The proposed method can detect the signal directly without any channel equalization methods in short range multi-path channel. The existing deep learning methods DBN and SAE can be applied to our system. Meanwhile, we propose a novel deep learning method - TTN with a lower computational complexity compared with DBN and SAE. To evaluate the performance of the proposed system, series of comprehensive simulation experiments is conducted under the environment of multi-path channels. The experimental results show that the proposed deep learning detection method can be used for signal demodulation in multi-path channel without channel equalization.