Structured covariance matrix estimation for the range-dependent problem in STAP

We propose a method to compute an estimate of the clutter-plus-noise covariance matrix in bistatic radar configurations. The estimation is based on the computation of the clutter scattering coefficients based on a single data snapshot at each range using a model of the received signal. The covariance matrix of the data is modeled as a structured covariance matrix with the scattering coefficients as unknown parameters. The method is based on the computation of the maximum likelihood. We use the Expectation-Maximization method as estimation benchmark. Since the problem is ill-posed, regularization is mandatory. This regularization is performed by spatial smoothing. The method we propose, unlike the Expectation Maximization, is not iterative and is thus less computationally demanding. The obtained covariance matrix estimate is used to compute the matched filter in order to perform target detection. The performance of the proposed estimation method is evaluated in terms of signal to interference-plus-noise ratio (SINR) losses and is found to be almost indistinguishable from the performance of the clairvoyant case.