Parameter estimation and elimination for stochastic noise from Fiber Optical Gyro in wavelet domain

Stochastic noise of fiber optical gyro (FOG) mainly appears as white noise (WN) and 1/f^gamma fractal noise. Based on the calculation and analysis of the auto-correlation function of sampling data from an interferometric FOG, it can estimate the variance of WN. Then according to the properties of fractal noise in wavelet domain, it can figured out the parameters of fractal noise through fitting practical variances, which are the results of subtracting the variance of WN from that of wavelet coefficients at each scale, by an optimum exponential function. To get a high SNR signal, it selects an improved threshold-value of wavelet filter. The experiment result shows that it achieved a better effect, even if the distribute of white noise do not know beforehand.