Unlike the Brillouin scattering, the anti-Stokes Raman scattering in optic fiber is unrelated with the strain, but is only the function of the absolute temperature. The frequency shift caused by Raman scattering is about 13.95 Thz. So the Raman scattering is easier to be picked up than Brillouin scattering. It has certain advantage while being used as the signal of the distributed optic-fiber temperature sensor. But it is weaker than the Brillouin scattering, the peak photocurrent produced in APD is of the order of nA,near or even lower than the noise current of the APD. The N/S after being photo-electric transformed is usually lower than 1. The means of Fourier transform and wavelet transform is not effective in dealing with such kind of signal. Through analyzing the feature of the anti-Stokes Raman scattering signal and using the wavelet, the anti-Stokes Raman scattering signal after being cumulated & averaged was picked up. The supersensitive detecting below the “noise current” of the APD was carried out with the senstivity: 0.104 nA·K-1,lower than the “noise current” of the APD 2 classes.