Improvement of ICA in Noisy Environment

H. Nakadaira and S. Wada (Japan)


Adaptive Filters, Signal Noise Control,Independent Component Analysis


In this paper, a robust ICA algorithm in noisy environment is investigated to improve the convergence and separation capabilities of the basic ICA algorithm. First, in order to improve the convergence characteristics, a parameter used in the basic ICA algorithm is selected as a time function of slowly decreased scheme. Then the scaling process is applied to the elements of iterated matrix. Further, a generalized spectrum subtraction method is used to reduce the effect of noise and to improve the signal source separability of the algorithm. The effectiveness of improvement of the method is demonstrated with several signals in the simulations.

Important Links:

Go Back