On the Convenience of Perfect Channel Knowledge for Spatial Equalization of Correlated MIMO-VLC Links: Is it Really Worth it?

Visible Light Communication has risen as a promising technology to support indoor wireless connectivity and provide illumination services as well. Moreover, high communication performance can be achieved by implementing Multiple-Input Multiple-Output architectures. However, the crosstalk characterizing spatially correlated channels scenarios may significantly impact on reliability. Such issue is commonly addressed by means of spatial equalization, with zero forcing and minimum mean square error representing the most widespread approaches. These methods allow the mitigation of crosstalk, even though they may cause undesired noise amplification as side effect, inducing errors during signal detection and decoding. In this regard, by focusing on a more realistic approach with respect to the literature, this work overturns the belief that having perfect channel knowledge is always better. Hence, we show that, when applying spatial equalization, imperfect channel knowledge is sometimes preferable to ideal channel state information. As a result, noise amplification is reduced at the expense of a reasonably less accurate crosstalk mitigation, with better communication performance being achieved.