Title :
A new Blind Source Separation algorithm based on non-orthogonal joint diagonalization of second-order statistics under lower SNR
Author :
Wang Ping ; Chen Jian ; Lu Ji xiang ; Wang Xin
Author_Institution :
Sch. of Telecommun. Eng., Xidian Univ., Xi´an, China
Abstract :
To solve the problem of Blind Source Separation (BSS) of the communication signals without knowing the source number, we propose a new algorithm based on the second-order statistics. By jointly diagonalizing the time delay correlation matrix of the observed signals and using the improved new non-orthogonal joint diagonalization (NOJD), a better solution of BSS under lower SNR can be achieved. The simulation result shows that the proposed algorithm can successfully separate communication signals with SNR as low as 10dB under conditions such as the source number is unknown or dynamically changing, and in the over-determined mode regardless of the signals´ modulation methods. We use Signal to Interference Ratio (SIR), Crosstalk Error(CTE) and Correlation Coefficient as the performance indexes to prove the superiority of the proposed algorithm over the classical Second-Order Blind Identification (SOBI).
Keywords :
blind source separation; correlation methods; statistical analysis; blind source separation algorithm; correlation coefficient; crosstalk error; lower SNR; non-orthogonal joint diagonalization; second-order blind identification; second-order statistics; time delay correlation matrix; Correlation; Delay effects; Jacobian matrices; Joints; Signal processing algorithms; Signal to noise ratio; Source separation; Blind Source Separation; Correlation Coefficient; Crosstalk Error; Non-orthogonal Joint Diagonalization(NOJD); Over-determined; SIR; component;
Conference_Titel :
Mechatronics and Automation (ICMA), 2010 International Conference on
Conference_Location :
Xi´an
Print_ISBN :
978-1-4244-5140-1
Electronic_ISBN :
2152-7431
DOI :
10.1109/ICMA.2010.5588452