Research of Denoise Technique for 1553B Bus Based on Wavelet Transform

摘要

MIL-STD-1553B bus has been the most popular militarized network so far. The main elements affecting the performance of data bus are Gaussian noise introduced from transmission and impulse noise. When data bus is contaminated by noise, the signal waveform will be extended or shrank which results in U-turn of the rise or fall process, and so on. The traditional denoising method for data bus is low-pass filter, which removes high frequency component mainly caused by noise, at the same time, the singularities and irregular structure represented by high frequency component that often carry the most important information are all removed. In this paper, a novel method based on a local estimation of the signal regularity in wavelet domain is proposed to reduce Gaussian noise. We develop an algorithm that discriminates a signal from Gaussian white noise by analyzing the behavior of the wavelet transform local maxima that describe the singularity of signals and noise. Compared with classical low-pass filter, our method has better denoising performance. Moreover, since we suppress most of wavelet coefficients mainly containing noise at the finest scale, the computation complexity is reduced greatly compared with classical modulus maxima method.