Existing blind equalizers employing the least mean squares (LMS) principle suffer from slow convergence rates. An obvious solution is to replace the LMS-type algorithm used in a blind equalizer with a `fast algorithm' that is known to have a faster convergence speed in non-blind cases. Unfortunately, such a direct replacement cannot guarantee convergence in a blind environment. This paper offers a solution by exploiting an analogy between the self-orthogonalizing (multidimensional Newton) method and the fast transversal filter (FTF) algorithm. The resulting blind equalization algorithm has a structure similar to that of the FTF algorithm and, as numerical examples show, yields convergence rates faster than those of LMS-type blind algorithms. The computing complexity is linearly proportional to the number of taps of the associated transversal filter, just as with the FTF algorithm. We also propose a new error signal and present the associated convergence analysis.
|頁（從 - 到）
|Journal of the Chinese Institute of Electrical Engineering, Transactions of the Chinese Institute of Engineers, Series E/Chung KuoTien Chi Kung Chieng Hsueh K'an
|Published - 1 5月 1996