Identification of impulse responses of multi-input systems

This paper considers the identification of impulse responses of systems with multiple inputs. An existing technique utilizing signal sets which are correlated with one another is modified to improve its convergence properties. A detailed example is presented to compare the performance between identification using a set of correlated signals and identification using a set of uncorrelated signals. In particular, it is shown that correlated design has the capability to outperform uncorrelated design in terms of estimation accuracy, signal design complexity and experimentation time.