Dictionary Matrix Map in Massive MIMO Systems

As the number of base station (BS) antennas increases, conventional massive multiple-input multiple-output (MIMO) channel estimation methods based on the orthogonal pilots become infeasible. To address this issue, channel knowledge map (CKM) has been proposed to enable environment-aware channel estimation. In this paper, we propose to construct a particular type of CKM, namely dictionary matrix map (DMM), based on the dictionary matrices. Specifically, the coverage of the target BS is divided into multiple regions and the DMM is constructed by storing one particular dictionary matrix for each region. The DMM can be constructed based a predefined dictionary matrix or learned dictionary matrices. Based on the DMM, the DMM-aware least square (LS) estimator and compressive sensing (CS) estimator are proposed to reduce the pilot contamination due to the pilot reuse. Simulations demonstrate that the proposed DMM-aware channel estimators can significantly improve the channel estimation performance.