Abstract—This paper proposes a new cyclic prefix (CP) based extended QR decomposition Recursive Least Squares (QRD-RLS) algorithm using block Householder transformation (HT) for adaptive channel estimation in multicarrier systems. The new method is computationally cheaper than CP exploiting correlation matrix based block RLS (CMB-RLS) adaptive channel estimation approach, and possesses desirable properties for systolic array implementation. Simulation results suggest that as compared to the CMB-RLS scheme, the new algorithm possesses superior steady state performance.