Source Identification Using Multi-Input/Single-Output Modeling and Causality Checking of Correlated Inputs

This paper presents experimental results of a case study of source identification using multiple-input/single-output modeling in a case where the inputs are coherent to some extent and, hence, the priority among the correlated inputs must be decided before applying the partial coherence function approach. The basic idea is that either one of any two correlated signals causes the other and that this causality can be checked by observing the impulse response functions estimated in the negative time region, interpretations of which are provided for a system transfer function given in the fractional form of polynomials and for a case of wave propagation. The experimental results from a three inputs/single output acoustical system shows that the method works well and is promising in the source identification problems with coherent inputs.