MIMO-OFDM channel detection algorithm in multi-station and multi-satellite uplink system based on deep learning

In the Earth-star uplink system composed of multiple measurement and control stations and multiple satellites, the multipath effect caused by cloud reflection or refraction and the Doppler effect generated by the relative motion between the measurement and control station and the satellite will cause the injection signal distortion and attenuation. This paper proposes a channel estimation algorithm based on deep learning theory, which is different from the traditional orthogonal frequency-division multiplexing (OFDM) system in which the channel state information is estimated first. And then the estimated channel state information (CSI) is used to design the equalization module. In the process of restoring the transmission signal, this paper designs and uses a five-layer fully connected neural network to restore the useful information transmitted by the measurement and control station to the satellite in an end-to-end manner. The simulation results show that the channel estimation algorithm based on deep learning can effectively reduce the use of pilots, submit spectrum utilization, and reduce the system error rate. When pilots with the same length as the signal are used in OFDM symbols, the bit error rate performance of channel estimation algorithms based on deep learning is better than the traditional Least Squares (LS) algorithm. The performance of the LMMSE (Linear Minimum Mean Square Error) algorithm is equivalent to the proposed algorithm also in terms of bit error rate. When the pilot sequence length is reduced to 1/8 of the OFDM symbol, the algorithm proposed in this paper is far superior to the traditional Channel estimation algorithm. This result can be seen that the algorithm can greatly improve the spectral efficiency by reducing the use of pilots.