Asymptotically optimal methods in a problem of the fastest detection of a change point .2. Studies of fastest detection methods

Studies are made of the asymptotic optimality properties of fastest change-point detection methods, namely, the cumulative sum (CUSUM) method and the Girshik-Rubin-Shiryaev (GRSh) method of an averaged likelihood radio for dependent random sequences. An, asymptotically optimal method of the ''moving-window'' of observations is suggested and a comparative analysis of the quality characteristics of this method and those of the CUSUM anti GRSh methods is carried out. Robustness properties of the CUSUM and GRSh fastest detection methods are investigated and results of the simulation modeling are listed.