Efficient Algorithms for Robust Factored Radar STAP

We deal with robust adaptive beamforming (RAB) for factored radar space-time adaptive processing (STAP) as a worst-case signal-to-interference-plus-noise ratio (SINR) maximization, because of mismatches in the desired signal and the interference-plus-noise covariance. To this end, we first reformulate the (beamformer unstructured) RAB problem as a second-order cone program (SOCP), and forcing the factored structure in the STAP weight vector, we come up with a new problem solved via an alternating optimization method. Second, we transform the worst-case SINR maximization into a biquadratic matrix inequality (BQMI) problem. We show that there is a closed-form solution for the BQMI problem, which implies that the robust optimal choice for the factored STAP processor is the Kronecker product of the spatial and temporal steering vectors. This is the most important finding of this paper suggesting that in the presence of steering and covariance mismatches (following the considered model), it is convenient to avoid adaptivity and to just use the non-adaptive beamformer.