Channel Estimation Based on DFT Interpolation and Out-Of-Band Spatial Information in Multi-User Millimeter Wave Systems

Millimeter wave (mmWave) MIMO communication can greatly increase the data rate, but the huge overhead of channel estimation cannot be ignored. In order to reduce training overhead, some researches based on the spatial channel correlation between the sub-6 GHz band and the mmWave band have been carried out. However, their performance is not satisfactory in distinguishing two users with close angular distance. In this paper, we propose a novel channel estimation method based on out-of-band information to enhance the accuracy of channel estimation in multiuser scenario. More specifically, based on the observation that dominant paths in mmWave band are a subset of that in sub-6 GHz band, we first obtain a coarse estimate of the spatial angle information from the sub-6 GHz band, and then, we use the discrete Fourier transform (DFT) interpolation method to refine the the estimation in mmWave band with low the training overhead. Numerical results show that, compared with the existing out-of-band information based channel estimation methods, the proposed method has better performance in multiuser scenario and the training overhead can also be reduced.