To reduce the computational complexity of block diagonalization precoding algorithm and ensure the BER performance,a block diagonalization precoding algorithm，named as LDL^H-BD(QR)，was proposed for multiuser multi-input multi-output (MIMO) systems，based on pseudo inverse and LDL^H decomposition.Firstly,computational complexity was reduced by combining global channel matrix pseudo-inverse and QR decomposition. Therefore,channel block diagonalization was fast finished. Furthermore,the LDL^H decomposition which has moderate subchannel gains uniformity was applied to optimize the equivalent channel for each single-user,so that procoding matrix was completed. Simulation results showed that the proposed algorithm has less computational complexity and similar bit error rate performance as the traditional block diagonalization algorithm.
