Modern analysis procedures for multiple input/output problems

This paper reviews the basis for new efficient data analysis procedures to solve multiple input/output problems. Practical iterative computational techniques are explained for obtaining conditioned spectral density functions, partial and multiple coherence functions, and other desired quantitites. Examples are given on how to identify different multiple correlated noise sources and how to determine their separate frequency response functions to any desired response point.