MIMO downlink communication employing multi-user transmitter pre-processing based on vector quantized channel spatial information: Performance results

Abstract

In this letter, we report the performance of multi-user transmitter pre-processing (MUTP) assisted multiple-input multiple-output downlink (DL) communication, when the channel state information (CSI) required to formulate the preprocessing matrix is estimated at the receiver and fed back to the base station (BS) through feedback channels that experience noise. In particular, in our work the CSIs are estimated at the mobile stations (MSs) and the estimated CSIs (ECSIs) are decomposed by invoking singular value decomposition. The signal space of right-hand side unitary matrix of the decomposed ECSI associated with each of the MSs is then vector quantized and the magnitudes and phases are communicated to the BS as channel spatial information through noisy feedback channels. This vector quantized channel spatial information which is tainted by noise is recovered by employing minimum mean square error based linear detector. The recovered spatial information is then utilized to conceive the pre-processing matrix to deal with the DL multi-user interference (MUI). Our study shows that, MUTP realized with perfect CSI at the BS is capable of completely eliminating the MUI. However, vector quantized channel spatial information based MUTP results in imperfect removal of MUI, as the quantization errors and feedback channel induced errors play a principal role in determining its performance in the context of interference removal. Albeit the achievable symbol error rate (SER) slightly degrades compared to the perfect CSI case, we advocate that vector quantization seems to be an efficient approach in quantizing the necessary spatial information and feeding them back to the BS for the purpose of formulating the pre-processing matrix particularly in frequency division duplex aided wireless systems.