Blind Separation of Weak Signals under the Chaotic Background

In the paper, to solve the problem that some existing methods of separating the weak signals from mixed chaotic signals have to use certain priori knowledge of chaotic signals such as the inherent properties, a FastICA method based on the negentropy is employed to separate the weak signals from the unknown mixed chaotic signals blindly. According to the maximum nongaussianity which is one of the basic ICA estimation principles, the algorithm uses negentropy as the measure. Then, the independence and high-order statistics information of every source of mixed chaotic signals are fully utilized, and a better separation performance can be obtained. The simulation results indicate that the weak signals can be separated fast and effectively and the error is relative less, even when the simulation is under the low SNR as -87.6 dB.