A new generalized cross correlator

Abstract

A new generalized cross correlator (GCC) for the passive time delay estimation problem is presented. The interpretation of this GCC is that of estimating the cross-correlation function by cross correlating the least mean-square estimates of the signal component in each of the observed waveforms. The implementation is simply a GCC with the weighting filter equal to the magnitude coherency squared. Numerical evaluation of the performance of this processor and a robust version indicate that they compare favorably to some of the well-known GCC procedures.