Tensor-Based Joint Channel Estimation and Symbol Detection for Time-Varying mmWave Massive MIMO Systems

In this paper, a tensor-based joint channel parameter estimation and information symbol detection scheme is developed for millimeter wave (mmWave) massive multiple-input multiple-output (MIMO) communication systems. At the base station (BS), the information symbols are encoded according to the Khatri-Rao space-time (KRST) method and transmitted through time-varying channels. The received signals at the mobile station (MS) are constructed into a nested complex-valued parallel factor (PARAFAC) model, which contains an outer model and an inner model, respectively. With outer model, we estimate the compound channel matrix and detect the information symbols considering the sparse scattering nature of mmWave channels. With inner model, we extract physical parameters, including angles of arrival/departure (AoAs/AoDs), Doppler shifts and complex path gains from the estimated compound channel matrix. These physical parameters can be used to significantly reduce feedback overhead. A tricky way here is that we convert complex inner-submodel into a real one such that the computational complexity is reduced. Compared with existing schemes, the proposed one improves the estimation accuracy with low computational complexity, and is applicable for both uniform linear arrays (ULAs) and uniform planar arrays (UPAs). Simulation results demonstrate the effectiveness of the proposed joint channel estimation and information symbol detection scheme.