Modelling and equalization of rapidly fading channels

Abstract

Basis expansion ideas are employed in order to equalize frequency-selective, rapidly fading channels. For certain multipath fading channels (e.g. the mobile radio channel), the time variations of the coefficients can be modelled as a combination of a small number of complex exponentials, under the assumption of linearly changing path delays. Based on this observation, novel adaptive and decision feedback algorithms are derived which exploit such an explicit modelling of the channel's variations. The problem of estimating the frequencies of the exponentials is also addressed using second- and higher-order cyclic statistics. By integrating ideas from cyclostationary signal analysis, both batch and recursive methods are developed. Finally, some illustrative simulations are presented.