Geodesical Refinement of MIMO Limited Feedback

In this paper, we investigate a codeword refinement technique called double codeword coding in the context of MIMO feedback. In the proposed method, the second codeword is fed back in addition to the best codeword as their interpolation at the transmitter improves the channel knowledge. 

The derivation of the optimal interpolation parameter is presented together with the derivation of an upper-bound to the interpolation parameter based on channel quality feedback available at the transmitter. We show that double codeword feedback refinement has slightly lower complexity than alternative techniques and still achieves similar performance. 

Furthermore, double codeword coding can be generalized to higher transmission ranks. Using extended multiuser MIMO link simulations, we verify that the double codeword feedback is able to provide significant gains in 3GPP LTE context.