Extraction of FOG signal from fractal noise

In a fiber optic gyro (FOG), there exist fractal noises characterized by non-stationary, self-similarity and 1/f/sup /spl gamma//spectral density. It is difficult to eliminate fractal noises using traditional methods. Based on the wavelet transformation, multi-scale wavelet decomposition is carried out so as to make the received signals become similar stationary signals. Some non-stationary properties like self-similarity and long-term dependency of fractal signals are attenuated in each channel. Multi-scale whitening filters are provided as well as the condition for whitening. So that the FIR filters can be applied to obtain FOG signals polluted by noises. The simulations results demonstrate that this method is simple and efficient to eliminate fractal noises from a fiber optic gyro. Compared with the de-noising by soft-thresholding in wavelet domain, the signal is restored better and the fractal noises are removed effectively.