A novel codebook design with the LBG algorithm in precoding systems under spatial correlated channel

Abstract

Precoding is an efficient approach to obtain high channel capacities and high quality in multiple-input-multiple-output systems and draws much attention in recent researches. In the codebook-based precoding systems, the tradeoff between performance and feedback overhead should be considered. In this paper, a novel version of the Linde-Buzo-Gray (LBG) algorithm is first utilized in the codebook design by redefining a set of operations in the conventional LBG algorithm. Moreover, the codebook can be trained online under spatial correlated channel adapting to the channel variation. Simulation results demonstrate that the proposed codebook design achieves better robustness, smaller quantization error and higher capacity than the traditional fixed uniform codebook design does.