Optimal Quantization Schemes for Orthogonal Random Beamforming - A Cross-Layer Approach

Orthogonal random beamforming (ORB) has recently attracted significant interest because of its ability to exploit multi-user diversity and spatial multiplexing gains by exclusively using partial channel state information (CSI). In this paper, we explore the impact of CSI quantization on the performance of ORB. By resorting to a dynamic programming formulation, we identify the optimal (i.e. pdf-matched) quantizer for which the sum-rate distortion is minimized. This is a cross-layer approach in the sense that the optimal quantizer (physical layer) actually depends on other layers´ parameters such as the number of admitted users (link layer). Performance is assessed by means of computer simulations and compared with that of a uniform quantizer.