Digital Compensation for Analog Front-Ends: A New Approach to Wireless Transceiver Design
by Prof. François Horlin, André Bourdoux
B
ML Channel Estimator
The channel equalization methods integrated in most of the digital communication systems usually rely on the knowledge of the channel impulse response. Pilot symbols are often interleaved in the data symbols based on which the channel impulse response can be estimated. A matrix model is typically built that links the channel impulse response vector h of size L to the observed vector of samples r of size Nr:
![]()
in which
of size Nr × L is a matrix composed of the transmitted pilot symbols and n of size Nr is the additive noise composed of independent Gaussian elements of variance
.
The ML estimate
of the channel impulse response vector h is selected such that:
![]()
![]()
where the second equality has been obtained by taking the expression of the noise probability into account. After optimization, the ML estimate is given by Deneire et al. (2003) and Kay (1993):
where ...