Quickest detection of superimposed hidden Markov models using a multiple target tracker

This paper addresses the quickest detection of superimposed hidden Markov Models (HMMs) using tracking algorithms. It is assumed that a known HMM is always extant but at an unknown time a second known HMM might be present and overlapped with the previous one. We develop a Page-like test to detect this superimposed HMM as quickly as possible. It is argued that such a scheme is feasible only for low dimension HMMs and therefore it is not extensible to multiple superimposed HMMs. A new detection method based on multiple target tracking ideas is presented, and it is shown through numerical examples that this scheme works essentially as well as the 'optimal' one.