Change-Point Detection in a Linear Stochastic System from Noisy Observations

We consider the problem of detecting the change point in the parameters of a discrete-time multivariate stochastic process described by stochastic difference equations. Only part of the process components are observable. A sequential change-point detection procedure is proposed. Formulas are derived for the procedure characteristics: mean time between false alarms and mean change-point detection delay. The asymptotic relationship between these characteristics is established.