Robust normalized adaptive time delay estimation approach based on zero order statistics
Wenhong Liu, Qiu Tian-shuang · Signal Processing · 2008
Based on zero-order statistics,the logarithmic moment processes,logarithmic wide sense stationary and ergodic random processes are defined.A robust normalized adaptive approach for time delay estimation (TDE),referred to as NZOSTDE,is proposed in the presence of an impulsive noise.Through a logarithmic transform,the impulsive signals without finite variance become the logarithmic- order signals.In this ease,all of the moments are finite.The NZOSTDE algorithm considers this property and the concept of geometric power,then the FIR filter models the two received signals added impulsive noises and its coefficients can be adaptive obtained under least mean square (LMS) error criterion in the logarithmic domain.Accordingly,the estimation of time delay between the two sources is the time index of the maximum in the coefficients.The proposed novel approach overcomes the shortcoming of algorithms based on frac- tional lower-order statistics (FLOS).Computer simulation results show that NZOSTDE algorithm is more robust than normalized least mean p norm time delay estimation,referred to as NLMPTDE,under very impulsive environments.