Geodesic Prediction for Limited Feedback Multiuser MIMO Systems in Temporally Correlated Channels

Limited feedback in multiuser multi-antenna systems is a practical technique for providing partial channel state information (CSI) to the transmitter. Unfortunately, it is known that the number of feedback bits per mobile terminal must be scaled with the number of transmit antennas and system signal to noise ratio. Therefore very high resolution codebooks are needed to approach the theoretically achievable sum rate. In this paper, we propose a geodesic prediction algorithm to 1) reduce the feedback overhead under temporally correlated channels, and 2) improve on the sum rate ceiling observed with a fixed codebook size. The proposed algorithm provides local refinement of the CSI thereby achieving higher resolution CSI at the transmitter despite of using a fixed codebook size. Furthermore, feedback can be performed infrequently to lower the average feedback rate. Numerical results are provided to illustrate the performance of the proposed algorithm.