Application of the denoising technique based on improved FastICA algorithm in cross-correlation time delay estimation

Abstract

Cross-correlation is a common time delay estimation method. However, in the actual measurement environment, the near-end and far-end sensors are often disturbed by the correlated noise, which seriously affects the accuracy of time delay estimation and even leads algorithm is proposed, and the steepest descent is introduced to reduce the sensitivity of FastICA algorithm to initial values. The useful acoustic data and disturbing acoustic data were measured respectively in a simulated granary. From these data, the signals contaminated by noise were generated by a mixing matrix and used for testing the performance of the proposed method. The testing results show that the proposed method can effectively suppress the influence of correlated noise on cross-correlation time delay estimation, and improve the accuracy of estimation.