Detection of a signal with unknown moments of appearance and disappearance (Q808095)

Let \(x_ 1,x_ 2,..\). be a sequence of independent r.v.'s such that during some unknown ``time-interval'' this sequence contains a signal. More precisely, if at time j there is a signal, then the density of \(x_ j\) is \(p_ 1\), otherwise \(x_ j\) has density \(p_ 0\). Here \(p_ 0(x)\) and \(p_ 1(x)\), \(x\in R^ 1\), are given probability density functions. It is assumed that both the moment of appearance and the moment of disappearance of the signal are unknown. Thus the problem is to detect the signal on the basis of the available observations. Two cases are treated, when the time for observations is finite, and when it is infinite. The author follows the adaptive Bayes approach in order to describe an algorithm as a solution of the above problem.