Prediction of Signal to Noise Ratio Based on Deep Learning in Tactical Wireless Communication Systems

Abstract

In this paper, we propose a prediction technique for future signal to noise ratio based on deep learning in tactical wireless communication systems. The communication system considered in this paper receives with multiple antennas and transmits in the future using the same antenna. We propose a deep learning model that predicts SNR for each transmit antenna in future transmission situations based on SNR received from multiple receive antennas in the past. Received ratio or a recorded ratio of a received SNR is set to 10 to 100 . If there is no received record nor SNR record, the received SNR is set through linear interpolation of the previously received SNR and the subsequent received SNR. According to the simulation results, wideband signals (4MHz) show better predictive performance than the narrowband signals (25kHz). In case of wideband, the proposed method is about 0.37  to 0.98  superior to the conventional method when the moving speed is over 20 . For narrowband signals, the proposed method is about 0.29  to 0.88  better than the conventional method for the moving speed over 20 . Those result indicates that the proposed prediction technique can be applied to antenna selection problem that provides the best SNR.