The Space-time Adaptive Processing Method with Generalized Sidelobe Canceler Structure by Using the Lp-norm Regularization

A method that further exploits the sparsity of weight vector in space-time adaptive processing (STAP) and decreases the training sample demanding is proposed. In the proposed method, STAP with the generalized sidelobe canceler (GSC) structure by using the L-p-norm regularization, instead of the conventional L-1-norm regularization, is considered. The core idea behind the method lies in minimizing the cost function that combines the output error and the L-p-norm of weight vector, which is a more accurate sparsity representing operator than L-1-norm. For solving the desired weight vector, the recursive least squares (RLS) algorithm with respect to the gradient of L-p-norm is deduced. In addition, since the diagonal loading factor cannot be clearly defined, the L-p-norm GSC STAP with parallel structure is designed by choosing the branch that maximizes the output signal to clutter ratio. In the end, the simulation experiments are conducted to verify the performance of the proposed L-p-norm GSC STAP.