FIR system identification using fourth-order cumulants with application to channel equalization
Boaz Porat, B. Friedlander · IEEE Transactions on Automatic Control · 1993
Three algorithms for finite impulse response (FIR) system identification from output measurements, using fourth-order cumulants, are described. The impulse response is assumed to be complex, and the input is a complex independent and identically distributed (i.i.d) non-Gaussian sequence. The algorithms are based on a certain low-rank matrix constructed from the output cumulants. An application of the algorithms to blind equalization of digital communication channels is discussed and illustrated by some examples.>