Tensor-Based Channel Estimation for Near-Field Millimeter Wave XL-MIMO Systems

In this letter, we propose a tensor-based channel estimation method for near-field millimeter-wave (mmWave) XL-MIMO systems. Specifically, we analyze the structure of the received space-time-frequency (STF) tensor at the mobile station (MS) and demonstrate that it adheres to a CANDECOMP/PARAFAC (CP) decomposition. We develop an alternating algorithm to estimate the factor matrices. Then, detailed uniqueness analysis reveal the suitability of tensor modeling techniques with the near-field channel modeling framework. A correlation-based method is subsequently applied to the estimated factor matrices to extract the parameters needed for channel reconstruction. Simulation results show that the proposed tensor-based near-field channel estimation algorithm achieves competitive estimation accuracy while maintaining a low pilot signaling overhead.