Performance of Pre-processing Schemes with Imperfect Channel State Information

Pre-processing techniques have several benefits when the Channel State Information (CSI) is perfect. In this work we investigate three linear pre-processing filters, assuming imperfect CSI caused by noise degradation and channel temporal variation respectively. Results indicate, that the Linear MMSE filter achieves the lowest Bit-error-rate (BER) and the highest SINR when the CSI is perfect, whereas the simple matched filter may be a good choice when the CSI is imperfect. Additionally the results give insight into the inherent trade-off between robustness against CSI imperfections and spatial focusing ability.