DocumentCode :
2613367
Title :
Use estimation of Performance Index to realize adaptive blind source separation
Author :
Zhang, Tianqi ; Li, Lizhong ; Zhang, Gang ; Gao, Chenqiang ; Hou, Ruiling
Author_Institution :
Chongqing Key Lab. of Signal & Inf. Process. (CQKLS&IP), Chongqing Univ. of Posts & Telecommun. (CQUPT), Chongqing, China
Volume :
5
fYear :
2011
fDate :
15-17 Oct. 2011
Firstpage :
2322
Lastpage :
2326
Abstract :
For adaptive blind source separation problem, by estimating the mixing matrix, then we can get the global transfer matrix, further we can obtain the estimation of PI (Performance Index), so we propose a new variable step-size blind adaptive algorithm based on EASI (Equivariant Adaptive Separation via Independence) algorithm. The accuracy of separation is determined by the global transfer matrix, that is to say, the step-size of EASI algorithm is adjusted by the accuracy of separation, which can be controlled by the estimation of PI. When the separation result with low precision, the step-size is biggish; on the contrary, accuracy increases, the step-size decreases. The simulation results show that the performance of the proposed algorithm is superior to the conventional EASI algorithm and has improved the separation result of the blind sources.
Keywords :
adaptive signal processing; blind source separation; matrix algebra; EASI algorithm; PI estimation; adaptive blind source separation; equivariant adaptive separation via independence algorithm; global transfer matrix; mixing matrix estimation; performance index estimation; variable step-size blind adaptive algorithm; Accuracy; Adaptive algorithms; Algorithm design and analysis; Blind source separation; Convergence; Performance analysis; Signal processing algorithms; BSS (blind source separation); EASI; Variable Step-Size;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Image and Signal Processing (CISP), 2011 4th International Congress on
Conference_Location :
Shanghai
Print_ISBN :
978-1-4244-9304-3
Type :
conf
DOI :
10.1109/CISP.2011.6100745
Filename :
6100745
Link To Document :
بازگشت