Determinant based codebook search for correlated massive MIMO channels

In this paper, we consider the complexity problem related to the selection of the optimal codeword matrix in large size codebooks used in massive MIMO systems. We propose a novel determinant based search method that makes use of the optimality of the statistical precoding in correlated MIMO channels. Codebook (CB) search complexity is reduced by focusing the search process to a smaller set of admissible codewords thus minimizing the computational complexity when performing the brute force scan for the best codeword. Simulation results show that our algorithm not only provides a good tradeoff between complexity reduction and performance, it also provides robustness against channel estimation errors in comparison to other hierarchal codebook search algorithms.