Blind signal extraction algorithm in overdetermined mixtures
超定条件下的盲信号提取算法

Based on the signal kurtosis, a blind signal extraction algorithm was proposed in over-determined mixtures. Converting the mixture matrix into a series of Givens matrix identifications, the algorithm can extract signals one by one from the mixtures. When the all unknown independent components were separated, the other components were redundant signals that can be regarded as copies of one or some independent components. All source signals were separated while the algorithm was accomplished. The algorithm is easy to implement with fast convergent speed. Computer simulation confirms its validity.