Research on de-noising method of fiber grating multiplexing network based on LCEEMD-LWT

Abstract

A de-noising algorithm combining local complementary ensemble empirical mode decomposition (LCEEMD) and lifting wavelet transform technology (LWT) is proposed to deal with the crosstalk noise buildup in high-capacity fiber grating multiplexing networks. Complementary ensemble empirical mode decomposition (CEEMD) serves for decompose this original spectral signal, and the normalization permutation en-tropy (NPE) is applied for identifying high-frequency nonlinear sequences in low-order intrinsic mode function to suppress the random noise. The filtering accuracy is improved by further decomposing the high-frequency intrinsic mode function with LWT. Finally, reconstruction of de-noised signal. The simulation results show that the LCEEMD-LWT method achieves better quality evaluation indexes than LCEEMD and EEMD-LWT when processing 1dBlow SNR spectral signal. To study the noise elimination capability of this devised method for nonlinear non-stationary semaphores, this de-trended fluctuation analysis (DFA) algorithm served for evaluating the results. It is indicted that this devised way has this highest fractal scaling index (2.114), which is better than other de-noising methods.