Sequential method of detecting change in random processes of recurrence type (Q1068779)

This paper deals with a discrete-time multivariate random process whose stochastic properties change at an unknown time. The process models before and after the change are assumed known. A sequential procedure for detecting the change is proposed. The procedure is based on the method of least squares and can assure given probabilities of false alarm and of missing the change by suitably choosing procedure parameters. A numerical comparison with a familiar algorithm is given.