Recursive least squares constant modulus algorithm based on the QR decomposition

Abstract

A novel QR-RLS constant modulus algorithm called QR-RLS-CMA is proposed. Its potential advantages include numerical stability, computational efficiency and a fast convergence rate. Simulations are performed to compare the convergence performance and the blind extracting ability of the proposed QR-RLS-CMA to the conventional SGD-CMA for adaptive CMA array. Results indicate that the QR-RLS-CMA has a much faster convergence rate than the SGD-CMA in the initial convergence phase. It illustrates the effectiveness of the proposed method.