Soft channel information for wireless communications

We present an overview of techniques recently elaborated to accurately represent channel state information when it is affected by large uncertainty (¿soft¿ channel information). In that context, point estimates of the channel states are replaced by probability densities optimally representing the knowledge about the channel state. Methods based on entropy maximization were used to derive several models of prior probability density functions for wireless flat-fading MIMO channels for various cases of a priori knowledge about the channel properties, including in particular arbitrary correlation matrices. It is foreseen that the importance of such techniques will grow with the advent of ¿agile¿ devices, operating in complex environments where accurate estimation of all channel parameters is out of question.