Co-channel Interference Cancellation Method Based on Deep Neural Network for LEO Satellite Systems

TLDR: This paper proposes a method based on deep neural network, which can accurately reconstruct the transmitted signal component coupled into the receiver, and then eliminate its interference to the weak GNSS signal, and provides a technical reference for signal broadcasting of LEO navigation augmentation.

Abstract: Low earth orbit (LEO) satellite navigation augmentation system has higher ground received signal power and faster Doppler change than Global Navigation Satellite System (GNSS), improving the positioning accuracy with occlusion and shortening the time of integer ambiguity-fixing. Aggressive frequency reuse reduces the complexity of user receiver implementation, but causing harsh co-channel interference at the meantime. The co-channel interference cancellation can alleviate the interference and improve orbit determination precision with receivers on board LEO satellites. Hence, this paper proposes a method based on deep neural network, which can accurately reconstruct the transmitted signal component coupled into the receiver, and then eliminate its interference to the weak GNSS signal. Finally, a simulation analysis is carried out with the BDS B1 frequency signal, which illustrates that this method can be adapted to different types of non-ideal channels. This method has good co-channel interference cancellation effect, and provides a technical reference for signal broadcasting of LEO navigation augmentation.