Sequential non-bayesian detection/isolation of abrupt changes with some applications
Proceedings of the X International Conference “System Identification and Control Problems”, SICPRO’15
19 – 31
The problem of sequential non-bayesian abrupt change detection and isolation is addressed in the paper. The goal is to discuss some typical criteria of optimality, several algorithms and to examine their statistical properties. We also discuss how to deal with nuisance parameters and how to integrate the nuisance parameters in the statistical decision-making process. To illustrate the theoretical results, the problem of navigation system integrity monitoring and the problem of volume anomaly monitoring in network traffic flows are considered.