Multi-parameter estimation for cognitive radar in compound Gaussian clutter

In this paper, we consider the problem of multi-parameter estimation in the presence of compound Gaussian clutter for cognitive radar using the variational Bayesian approach. The main advantage of variational Bayesian is that the estimation of multi-variate parameters is decomposed to multiple estimation of univariate parameters, thus enabling analytically tractable approximations. Numerical tests demonstrate that the proposed approach leads to improved estimation accuracy than the expectation maximization (EM) method, particularly in the case of non-Gaussian nonlinear models and a small sample size.