Regularized RLS Adaptive Algorithm with Conjugate Gradient Method

The recursive least-squares (RLS) family of adaptive algorithms is attractive for the identification of long unknown systems due to their high convergence speeds. Despite the fact that the combination with the conjugate gradient (CG) iterative method can further improve the tracking capabilities of the RLS, the namely RLS-CG is still vulnerable to high-level interference signals. This paper proposes a regularization method for the RLS-CG, which aims to considerably reduce the adaptive filter’s update process during the manifestation of the high-interference scenarios. The theoretical model is validated using simulations.