Title :
Sequential blind extraction of mixed source signals with guaranteed convergence
Author :
Liu, Derong ; Hu, Sanqing
Author_Institution :
Dept. of Electr. & Comput. Eng., Illinois Univ., Chicago, IL, USA
Abstract :
This work presents a gradient-based method for sequential blind extraction of mixed source signals. In our approach the number of unknowns to be determined is only 2(m - 1) and there is no need for computing any matrix inversion. Our algorithm is guaranteed to converge. In addition, the convergence speed of our algorithm can be improved by adjusting a parameter. Simulation results show the operation characteristic, the effectiveness of our method, and the advantages over the existing algorithm.
Keywords :
blind source separation; convergence; gradient methods; matrix algebra; convergence speed; gradient-based method; matrix inversion; mixed source signals; operation characteristic; sequential blind extraction; Blind source separation; Computational modeling; Convergence; Cost function; Image restoration; Independent component analysis; Newton method; Signal processing algorithms; Speech processing; Vectors;
Conference_Titel :
Circuits and Systems, 2004. ISCAS '04. Proceedings of the 2004 International Symposium on
Print_ISBN :
0-7803-8251-X
DOI :
10.1109/ISCAS.2004.1329449
