Title :
A forward/backward bootstrapped structure for blind separation of signals in a multi-channel dispersive environment
Author :
Dinç, Abdulkadir ; Bar-Ness, Yeheskel
Author_Institution :
Corporate Computer Systems, Holmdel, NJ, USA
Abstract :
A multidimensional adaptive algorithm with a forward/backward bootstrapped structure for a dispersive channel environment is proposed. It is an alternative multisignal separator where the loop-bandwidth of the signal separator structure and steady-state performance are crucial. It separates superimposed convolutive multiuncorrelated signals. The bootstrapped adaptive algorithm, which does not require a training sequence, employs the minimization of output signal correlations as optimization criteria. The control algorithm is set for the multidimensional case. The learning process of the 2-D signal separator using computer simulation is investigated and compared with that of the least mean square (LMS) algorithm for different cross-channel eigenvalue spreads.<>
Keywords :
adaptive filters; computer bootstrapping; correlation methods; eigenvalues and eigenfunctions; learning (artificial intelligence); multidimensional digital filters; telecommunication channels; telecommunications computer control; blind separation of signals; computer simulation; control algorithm; cross-channel eigenvalue spreads; dispersive channel; forward/backward bootstrapped structure; learning process; multidimensional adaptive algorithm; steady-state performance; superimposed convolutive multiuncorrelated signals;
Conference_Titel :
Acoustics, Speech, and Signal Processing, 1993. ICASSP-93., 1993 IEEE International Conference on
Conference_Location :
Minneapolis, MN, USA
Print_ISBN :
0-7803-7402-9
DOI :
10.1109/ICASSP.1993.319513
