Linear Gaussian Models in Target Tracking

This paper studies a unifying framework for linear Gaussian models used in target tracking. We propose a customized linear Gaussian methodology (Integrated Ornstein-Uhlenbeck in local horizontal dimensions and 2nd order Ornetein-Uhlenbeck in local vertical dimension) for context exploitation in long-term target prediction. We illustrate this approach with targets known to exhibit bounded altitude excursions and discuss both its advantages and limitations.1,2