A Polynomial Fourier Transform Based Channel Prediction Scheme for High-Mobility Massive MIMO Scenarios

Abstract

In wireless communication systems, high-mobility scenarios such as the Internet of vehicles and high-speed trains are becoming ever more important and yet challenging. The high mobility of users causes severe degradation of system performance. One of the crucial reasons is that the high-speed movement of users leads to serious channel aging, resulting in the issue of outdated Channel State Information (CSI) at the transmitter side. In particular, the assumption of a static Doppler frequency spectrum does not hold anymore. Regarding this, in this paper, a channel model that encompasses the characteristics of the time-varying Doppler spectrum is adopted, and the model parameters are solved through the Polynomial Fourier Transform (PFT) algorithm. However, there is a trade-off between the computational complexity and the PFT parameters' accuracy. Hence a Short-Time Fourier Transform (STFT) algorithm is used to pre-estimate parameters to reduce complexity. In addition, to reconstruct the sparse signal in the STFT domain, the compressive sensing algorithm based on orthogonal matching pursuit is utilized and achieves superior results. Simulations show that by adopting the COST2100 channel model, the average throughput of this scheme improves about 5% and 20%, respectively, compared with those produced by using the matrix completion method and the polynomial iteration method in the high mobility scenarios at 300 km/h.