Rotative Quantization using Adaptive Range for Temporally Correlated MIMO Channels

The performance of multiple-input multiple-output (MIMO) systems can be enhanced with the availability of channel direction information at the transmitter. One of the known techniques to feedback channel direction (eigenvector) information is parametrization using Givens rotations, which is also adopted in IEEE 802.11n standard. In temporally correlated MIMO channels, the feedback overhead can be significantly reduced by using differential quantization techniques such as adaptive delta modulation (ADM). However, the high rate of oscillation and bounded property of Givens rotations makes it difficult to track them using ADM. In this paper, a new method called Adaptive Rotative Quantization (ADRQ) is proposed to quantize Givens rotations in temporally correlated channels. ADRQ quantizes the rotation of eigenvectors between consecutive channel instances using a uniform quantization with an adaptive quantization range. While ADRQ requires the same feedback overhead as ADM, it is demonstrated that ADRQ has significant performance improvement over both ADM and the conventional non-differential quantization at all levels of temporal correlation. Moreover, simulations show that ADRQ converges relatively fast.