Estimation of multiple target trajectories with time varying amplitudes

This paper is concerned with the estimation of multiple constant velocity target trajectories in a low SNR environment. Each target trajectory is characterised by the initial position and velocity which are to be estimated. A major difficulty is that the target amplitudes are unknown and will in general be time varying. The approach used is to model the target amplitude as an autoregressive (AR) process. A maximum likelihood estimator is derived for the parameters of the AR process and the unknown target position and velocity using the expectation conditional maximisation (ECM) algorithm