Joint tracking of manoeuvring targets and classification of their manoeuvrability (Q1773807)

A mixed two-dimensional diffusion process with a varying local dispersion matrix is considered. This matrix is proposed to be a semi-Markov process with two states. The mixed process is being observed at time instants constituting a regular lattice. The observed meanings are independent normal random values with parameters of corresponding normal distributions depending on current observable points of the original process. A computational procedure is proposed to decide what class from given three classes the mixed process belongs to. These classes differ by their sojourn time distributions in the corresponding semi-Markov process, although all the distributions have the same first moment.