ABSTRACT
In this chapter, we generalize the quickest changepoint detection problem to the following multidecision detection-isolation problem. There are M “isolated” points/hypotheses associated with a change that has to be detected and identified (isolated) as soon as possible after the change occurs. Alternatively, there may be M populations that are either statistically identical or the change occurs in one of them at an unknown point in time. It is necessary to detect the change in distribution as soon as possible and indicate which hypothesis is true after a change occurs or which population is “corrupted.” Both the rate of false alarms and the misidentification (misisolation) rate should be controlled by given (usually low) levels. We propose certain multihypothesis detection-isolation procedures that asymptotically minimize the tradeoff between expected value of the detection lag and the false alarm/misisolation rates in the worst-case scenario. At the same time the corresponding sequential detection-isolation procedures are computationally simple and can be easily implemented on-line.
