Prediction based quantization and optimization for transmit beamforming MIMO with outdated channel direction information

In transmit beamforming multi-antenna systems, the representative quantized channel direction information (CDI) may become outdated due to the feedback delay. Grassmannian predictive coding (GPC) has been shown as an effective method to overcome this problem. However, the existing GPC algorithm is still of practical concern as the quantization resolution may not be ensured at relatively low feedback rate. In this paper, under the framework of GPC, we propose a novel transmit beamforming scheme to account for the feedback delay. By designing a new prediction based quantization function and optimizing the prediction, the forthcoming CDI can be accurately tracked. In addition, we analyze the achievable rate/througput gain of the proposed scheme over the memoryless quantization approach. Numerical results validate the effectiveness of our proposed encoding/decoding algorithm and the derived rate gain upper limit.