Journal of Jilin University Science Edition

Previous Articles     Next Articles

Adaptive QuasiNewton Algorithm for Blind Source Separation Based on Canonical Correlation Analysis

ZHANG Ruifen, YE Jimin   

  1. School of Mathematics and Statistics, Xidian University, Xi’an 710071, China
  • Received:2016-10-20 Online:2017-05-26 Published:2017-05-31
  • Contact: YE Jimin E-mail:jmye@mail.xidian.edu.cn

Abstract: By the analysis of classical canonical correlation analysis criterion, we proposed a new criterion for source signal extraction. Using the online quasiNewton iteration algorithm to solve the new criterion, and then a new blind source extraction algorithm based on canonical correlation analysis was obtained and blind source separation was realized. Theoretical analysis show that the unique global minimum point of the new source signal extraction criterion is the maximum point of classical canonical correlation analysis criterion. Simulation results show that the convergence speed the new algorithm is faster.

Key words:  blind source separation, canonical correlation analysis, quasiNewton iteration

CLC Number: 

  • O213