Performance Enhancement of Codebook-Based Beamforming Using Least Squares Approach

A codebook-based approach for channel estimation and beamforming affords the advantage of low complexity; however, it suffers from limited resolution of estimated directional angles of a channel owing to its finite size. To resolve this problem in codebook-based systems, this letter proposes a novel approach that employs the least squares (LS) method to estimate a directional angle from the measured beam gains of a few codewords, enhancing channel estimation performance. A nonlinear LS problem is formulated and solved using a grid search method. The impacts of grid resolution and number of codewords used on the LS estimation performance are investigated. Computer simulations demonstrate that the proposed scheme outperforms conventional schemes in terms of channel estimation errors and achieved beam gains. Furthermore, the proposed scheme requires a training overload of fewer or an equal number of codewords to measure beam gains to estimate a directional angle, compared with conventional schemes that seek to address the same problem of limited resolution in channel estimation.