On the design of interference alignment scheme for multi-user MIMO with limited feedback

Obtaining accurate global channel state information (CSI) is critical for interference alignment (IA), in particular for multiple-input multiple-output (MIMO) system with high dimensional space. With limited feedback, a quantization approach is proposed with manageable complexity to quantize the vectors from channel matrices separately. By characterizing the rate loss and bit scaling requirement we further show how the amount of feedback bits affects the numbers of transmit/receive antennas that should be used. Simulations illustrate with this quantization approach the system can attain considerable rate improvement with large number of feedback bits and optimal numbers of transmit/receive antennas.