Title :
Blind Source Extraction
Author_Institution :
Singapore Inst. of Manuf. Technol., Singapore
Abstract :
In this paper, we proposed blind source extraction (BSE) technique for signal overlapping problem. The BSE method consists of an additional cascaded nonlinear model, which assists in blind extraction and also reduces the remaining deflation noise. In order to perform simultaneous the nonlinearity estimation and source extraction, a nonlinear cost function is introduced which optimises for both the independence of sources and inverse of the nonlinearity. A rigorous analysis is provided and the existence and uniqueness conditions for the solutions are given. Simulation results are presented which illustrate the validity of the proposed approach.
Keywords :
blind source separation; blind source extraction; cascaded nonlinear model; nonlinear cost function; signal overlapping problem; Adaptive algorithm; Algorithm design and analysis; Analytical models; Automatic control; Manufacturing automation; Robot control; Robotics and automation; Sensor phenomena and characterization; Source separation; Statistics; Blind source separation; blind source extraction; cascaded nonlinear model; deflation; post-nonlinear model;
Conference_Titel :
Control, Automation, Robotics and Vision, 2008. ICARCV 2008. 10th International Conference on
Conference_Location :
Hanoi
Print_ISBN :
978-1-4244-2286-9
Electronic_ISBN :
978-1-4244-2287-6
DOI :
10.1109/ICARCV.2008.4795778