Deterministic Vs Stochastic Modeling of Clutter for Adaptive Radar Detection*

Recently a class of adaptive radar detectors was derived based on a data model with the clutter represented in the mean of the distribution [9]. This is a departure from the traditional radar data model, which assumes that the clutter and all other interference are modeled in the covariance matrix. This new model allows for a reduced-dimension covariance matrix estimation which is helpful in scenarios with limited sample support. This paper will study the impact of clutter modeling on the performance of detection algorithms under both the traditional and the new model.