Receiver Design for ICI-CSK System: A New Perspective Based on GRU Neural Network

In conventional chaotic communication schemes, the receivers mostly use energy detection methods. However, due to the limitations of this detection method, the initial condition index chaos shift keying (CSK) scheme cannot work over the multipath Rayleigh fading channel. To overcome this issue, a gated recurrent unit (GRU)-aided initial condition index CSK system, referred to as GRU-ICI-CSK system, is proposed in this letter. The proposed GRU-ICI-CSK scheme exploits the powerful learning and classification abilities of deep learning, which can achieve a high reliable bit-error-rate (BER) performance. Moreover, we analyse the complexity of the proposed scheme. Simulation results demonstrate that the proposed scheme can achieve superior BER performance compared to the conventional energy detection chaotic communication schemes over multipath Rayleigh fading channels. Besides, the network based on GRU of the proposed scheme has lower computational complexity and better BER performance compared to the state-of-the-art network based on long short-term memory. Thanks to these benefits, the proposed GRU-ICI-CSK scheme can be considered as an effective solution for low-power and short-range wireless communication.