Radar signal detection in non-Gaussian distributed clutter by Bayesian predictive densities

We present a coherent radar signal detection scheme in non-Gaussian distributed clutter and its simulation results. In this scheme the clutter is modeled by compound Gaussian distribution and unknown parameters, i.e. target amplitude and clutter, are estimated based on a posteriori distribution with a noninformative prior. Also a technique called Bayesian predictive densities is employed. In order to investigate the performance, we carried out the Monte Carlo simulation and its results are also compared with conventional detection schemes such as maximum likelihood and maximum a posteriori estimator. The simulation results show its usefulness.