End-to-End Optimization of Multiuser MIMO Systems Using Autoencoders with Bidirectional Channel Estimation
Machine Learning, Multiple-Input Multiple-Output (MIMO), Signal Detection, Channel Estimation.
The spectral efficiency gains introduced by multiuser MIMO systems render them relevant schemes for current and upcoming generations of mobile communication networks. Due to the intrinsic complexity of the mathematical models of these systems under realistic conditions and the interdependence between the processing steps of the transmitters and receivers, machine learning is an option that allows the complete system to be designed by training a noisy autoencoder. This paper proposes a neural network architecture for end-to-end optimization of a multiuser MIMO system. The performance of the system, measured in terms of symbol error rate, was compared to an M-PSK baseline with zero-forcing equalization and least-squares channel estimation. Simulations were performed using a Rayleigh fading channel model and the realistic 3GPP TR 38.901 model. A bidirectional channel estimator, based on the interpolation of sparse pilots, is proposed, reducing the control signaling to less than 3% in exchange for a fixed 10 ms delay. The results show that signicant gains can be achieved by applying the proposed model, but those vary with respect to the estimation errors during the pilot transmission times.