It was reported recently that chaos properties could be used to relieve inter-symbol interference caused by multipath propagation in chaos-based wireless communication system. Although there exists the optimal decoding threshold to theoretically eliminate the inter-symbol interference, its practical implementation is still a challenge due to the strong requirement to know the future symbols to be transmitted. To tackle this almost ‘impossible’ task, convolutional neural network with deep learning structure is proposed to predict future symbols based on the received signal, to further reduce inter-symbol interference and to obtain a better bit error rate performance. Due to the short time predictability of chaotic signal, the proposed method is able to predict short-term future symbols and get a better threshold suitable for the time-variant channel. The analytical bit error rate of the proposed method is derived. The contributions of the paper are as follows: firstly, a convolutional neural network with deep learning structure is proposed for the first time to predict the future symbols in the chaos baseband wireless communication system, which does not require much training in this important application; secondly, the future bits predicted by the trained convolutional neural network are used together with the past decoded bits to calculate more accurate decoding threshold compared with the existing methods, yielding a better bit error rate performance. Numerical simulations and experimental results validate the effectiveness of our theory and the superiority of the proposed method.
Bibliographical noteFunding Information:
This work was supported in part by Shaanxi Provincial Special Support Program for Science and Technology Innovation Leader. Dr Bai was supported in part by China Postdoctoral Science Foundation Funded Project (2020M673349), and Open Research Fund from Shaanxi Key Laboratory of Complex System Control and Intelligent Information Processing (2020CP02).