Title :
Blind channel shortening of ADSL channels with a single-channel linear predictor
Author :
Dalzell, W.G. ; Cowan, C.F.N.
Author_Institution :
Digital Commun. Res. Group, Queen´s Univ. of Belfast, Belfast, UK
fDate :
Aug. 29 2011-Sept. 2 2011
Abstract :
Equalization of non-stationary communication channels is preferably blind, to minimise overheads that reduce data throughput, and adaptive so as to track the channel characteristic. In this paper we consider the use of a single-channel linear predictor as an equalizer for multi-carrier modulated (MCM) signals. The MCM method with its cyclic-prefix has a level of tolerance for inter-symbol interference. We show that despite an intrinsic minimum-phase characteristic, the predictor can equalize mixed-phase ADSL test channels satisfactorily. We also show that by increasing the predictor delay, channel-shortening may be explicitly introduced, and that use of this version of the predictor improves equalization performance at higher channel noise levels. Minimising computation loads of blind equalization algorithms is also desirable. The predictor has significantly lower computational load than other blind equalization schemes, though at a cost of lower convergence speed. There is potential for faster adaptation while retaining a lower computational load than other schemes.
Keywords :
blind equalisers; digital subscriber lines; intersymbol interference; linear predictive coding; MCM signals; blind channel-shortening; blind equalization algorithms; channel characteristic; cyclic-prefix; intersymbol interference; intrinsic minimum-phase characteristic; mixed-phase ADSL test channels; multicarrier modulated signals; nonstationary communication channels; predictor delay; single-channel linear predictor; Blind equalizers; Correlation; Finite impulse response filters; Prediction algorithms; Predictive models; Signal to noise ratio;
Conference_Titel :
Signal Processing Conference, 2011 19th European
Conference_Location :
Barcelona