A novel fast algorithm for blind signal separation
Weidong Zhou, Lei Jia · 2003
The independent component analysis (ICA) for blind source separation is investigated in this paper. A contrast function is given based on maximum likelihood and mutual information theory. A fast iterative ICA algorithm is derived by optimizing the function. The method does not need to calculate the higher order statistics of signals, and has the property of second-order convergence. The experimental results for separation of simulation and real signals using the proposed algorithm is presented.