Comparison of various correlation-based techniques for the estimation of impulse responses

This paper provides a comparison between three correlation-based techniques, namely the direct approximation, deconvolution and direct inversion methods, for the estimation of impulse responses in both single-input and multi-input scenarios. The comparison is made from the perturbation signal design viewpoint, as this is an important step in system identification. Classes of broadband signals which are commonly used for identification are reviewed. The constraints of the three approaches in terms of the minimum input signal period and harmonic specifications are considered, along with discussion on computational complexity. Simulation examples are used to highlight the similarities and differences between these methods. The results have wide-ranging applications particularly in the area of control and communications.