Blind identification of non-minimum phase ARMA systems

This paper presents a second-order statistics based method for blind identification of non-minimum phase single-input–single-output (SISO) auto-regression moving-average (ARMA) systems. By holding the system input while sampling the system output at the normal rate, the SISO system is transformed into an equivalent single-input–multi-output (SIMO) ARMA model. Theoretical analysis is conducted to exploit the system auto-regressive information contained in the autocorrelation matrices of the over-sampled output and to derive expressions for constructive estimation of the ARMA system parameters. The developed systematic identification method has flexibility in choosing the over-sampling rate which can be as low as two. The effectiveness of the proposed method is demonstrated by simulation results.