Optimal Sequential Detection of Signals With Unknown Appearance and Disappearance Points in Time

The paper addresses a sequential changepoint detection problem, assuming that the duration of change may be finite and unknown. This problem is importance for many applications, e.g., signal image processing where signals appear disappear at unknown points in time or space. In contrast to conventional optimality criterion quickest requires minimization expected delay given average run length false alarm, we focus on reliable maximin maximizing minimal probability (or space) window local maximal alarm prescribed window. We show optimal procedure modified CUSUM procedure. then compare operating characteristics this with popular engineering Finite Moving Average (FMA) algorithm ordinary using Monte Carlo simulations, which typically later algorithms have almost same performance as one. At time, FMA has substantial advantage -- independence intensity signal, usually Finally, applied detecting faint streaks satellites optical images.