Utilizing multivariate autoregressive model to reveal internal dependencies in multichannel measurement data

A method based on the estimation of the multivariate autoregressive model (MAR-model) is introduced to analyze multichannel data. Because the MAR-model is a black-box model and can describe systems with feedback-loops, it table for the analysis of complex closed-loop multivariate systems. The authors identify the MAR-model and, based on the model, decompose the multichannel spectral matrix. The proposed method offers a new possibility to analyze systems of which there is no exact prior knowledge of internal structures