An improved system blind identification method based on second-order cyclostationary statistics and the properties of group delay, has been proposed. This is achieved by applying a correction to the estimated phase (by the spectral correlation density of the system output) for the poles, in the group delay domain. The results indicate a significant improvement in system blind identification, in terms of root mean square error. Depending upon the signal-to-noise ratio, the improvement in percentage normalized mean square error ranges between 20 and 50%.

Giridhar, P.V.S.Narasimhan, S.V.2026-02-05Improved sSadhana - Academy Proceedings in Engineering Sciences, 2000, 25, 2, pp. 85-962562499https://doi.org/10.1007/BF02703751https://idr.nitk.ac.in/handle/123456789/280262000Decision support systemsFunctionsSignal to noise ratioSpectrum analysisStatistical methodsSystem theoryBlind identificationGroup delay approachSecond order cyclostationary statisticsIdentification (control systems)An improved system blind identification method based on second-order cyclostationary statistics and the properties of group delay, has been proposed. This is achieved by applying a correction to the estimated phase (by the spectral correlation density of the system output) for the poles, in the group delay domain. The results indicate a significant improvement in system blind identification, in terms of root mean square error. Depending upon the signal-to-noise ratio, the improvement in percentage normalized mean square error ranges between 20 and 50%.