IP Aided OMP Based Channel Estimation for Millimeter Wave Massive MIMO Communication

Adoption of hybrid precoding is the key element of reducing the hardware cost of radio-frequency compared with the conventional full-digital precoding approach in millimeter wave (mmWave) MIMO systems. For hybrid precoding, channel state information (CSI) is needed. However, the use of analog precoding in hybrid architecture and the large antenna array make channel estimation difficult for mmWave system. It has been shown that mmWave channels exhibit sparsity, thus compressive sensing (CS) techniques can be leveraged to conduct channel estimation. Conventional CS based channel estimation methods for mmWave MIMO are based on quantized angle grid. However, the performance would be severely affected by off grid angles which can not be improved by increasing the resolution because this will increase the coherence between the grid points. In this paper, we propose an Interior Point (IP) aided orthogonal matching pursuit (OMP) algorithm. It significantly improves the channel estimation accuracy by reducing the estimation error of angle-of-departure (AoD) and angle-of-arrival (AoA). The simulation results demonstrate the advantage of the proposed IP-OMP over the existing methods such as least squares and the conventional OMP.