| This thesis studies multi-antenna (also known as multi-input multi-output, or MIMO) communications with finite rate feedback from the receiver to the transmitter. In a point-point MIMO system, feedback enables transmission along the best spatial eigen-channels. In a multi-user/network system, feedback accommodates coordination among users/antennas to intelligently select good channels and reduce interference. Simulations demonstrate that a few bits of feedback can improve system throughput significantly.; The major thrust of this thesis is to understand the effect of finite rate feedback on throughput. On the theoretical front, analysis of finite rate feedback is closely related to differential geometry, specifically the Grassmann manifold, and random matrix theory (RMT). We study the quantization problem in the Grassmann manifold and derive the asymptotic rate-distortion tradeoff in several asymptotic regions. A synergistic combination of the derived results and RMT opens the door to a deeper understanding of the quantization of channel matrices. On the more directly applied side, we study many communication systems, including point-to-point, CDMA, multi-access and broadcast systems with finite rate feedback. Performance analysis is conducted and efficient designs are proposed. The developed schemes can be applied to many modern communication systems including 3GPP, WLAN and WiMax. |
