Abstract.
Given that no false alarm was made, the CUSUM process behaves like a stationary process. The corresponding transition density is derived for Erlang distributed random variables. Then a representation for the so-called average delay and a geometric approximation for the CUSUM run length distribution is obtained.
Similar content being viewed by others
Author information
Authors and Affiliations
Additional information
Received May 1997
Rights and permissions
About this article
Cite this article
Knoth, S. Quasi-stationarity of CUSUM schemes for Erlang distributions. Metrika 48, 31–48 (1998). https://doi.org/10.1007/s001840050003
Issue Date:
DOI: https://doi.org/10.1007/s001840050003