Blind identification of non-minimum phase ARMA systems

This paper presents a blind identification algorithm for non-minimum phase single-input single-output (SISO) plants using an over-sampling technique with each input symbol lasting for several sampling periods. First, an SISO autoregressive moving average (ARMA) plant is converted into its associated single-input multi-output (SIMO) system by holding the system input and over-sampling the system output. A sufficient and necessary condition for coprime transfer functions of the SIMO system is provided. A new second-order statistics (SOS) based blind identification algorithm for the SIMO ARMA model is then presented, which exploits the dynamical autoregressive information of the model contained in the autocorrelation matrices of the system outputs. Further, the transfer function of the SISO system is recovered from its associated SIMO transfer functions. Finally, the effectiveness of the proposed algorithm is demonstrated by simulation results.