An FK-transform based MU-MIMO precoding with low complexity

This signal to leakage and noise ratio (SLNR) based precoding algorithm using generalized singular value decomposition (GSVD) outperforms conventional SLNR based precoding algorithm in terms of numerical stability, but its computational complexity has proved to be high due to the necessary singular value decomposition (SVD) operation in the generation of precoder. In order to tackle this problem, a novel SLNR comprising of a precoding algorithm has proposed utilizing the Fukunaga-Koontz (FK) transform. The proposed algorithm also has the advantage of numerical stability, but without SVD operation, consequently lower computational complexity was discovered as to that of SLNR based precoding using GSVD. Theoretical analysis display the numerical stability of the proposed precoding by showing equivalency to the SLNR based precoding using GSVD. Meanwhile, the complexity was lower than the GSVD based precoding algorithm by a factor of 34.3%~42%, which makes the proposed precoding algorithm more suitable for practical implementation. Simulation results also demonstrate effectiveness and efficiency of the proposed precoding algorithm.