A Combined Codebook Constructing Method for Beamforming on Correlated MIMO Channels

Abstract

A combined codebook constructing method is proposed for beamforming in correlated Multiple-input multiple-output (MIMO) channels, where the codebook is dynamically constructed based on the transmit covariance matrix of the channel. Monte Carlo simulations show performance close to optimal beamforming. Huffman coding is used to further reduce the feedback rate in correlated MIMO system.

Adaptive Codebook Selection for Limited Feedback MIMO Beamforming Systems

“…Especially in AR(1) time-correlated channel, the adaptive codebook of 4 bit feedback is similar to that of 5 bit grassmannian codebook. The work is not compared with other methods such as in [3,9,10], etc. Because in spatially correlated channel, the adaptive codebook mentioned here is actually as in [10], which considers the disturbance of the channel, we can select only one child codebook according to the direction of the channel as the codebook to do quantization.…”

“…In this section we extend the work in [10] to multiple combined codebooks with different engen-directions (here we call the right singular vector corresponding to the maximum singular value of a channel matrix as eigen-direction). Each codebook corresponding to an engen-direction, and each time, system selects a codebook according to its eigen-direction to quantize the channel.…”

“…By rotating the Grassmannian line packing beamformers by the square root of the transmit correlation matrix, [9] obtains a beamforming codebook design criterion for correlated channel, and shows that receive correlation does not affect the distribution of the optimal beamforming vector. This method is extended to a correlated channel with disturbance [10], and there they propose a combined codebook construction method to deal with channel disturbance. They, however, do not consider time-varying correlation and the burden of feedback of the correlation matrices.…”

“…In [10], we use a combined codebook generation method to deal with spatially correlated channel which experience some disturbance. However, it does not consider time-varying correlation and the burden of feedback of the correlation matrices.…”

“…It can be useful when the channel existing only one direction or changing very slowly that its direction can be estimated. In this paper, we extend the method in [10] to a time-varying correlated channel, and propose an adaptive quantization structure based on the combination of two grade codebooks, where the two codebooks are combined together to make a set of children codebooks, and from which to choose the codebook for quantification. The codebook selection information is implied in the feedback codeword indexes, thus to switch the codebook adaptively.…”