Modified bias compensation recursive least-squares method for noisy FIR adaptive filtering

In this paper a modified bias compensation recursive least-squares (MBCRLS) method is proposed to deal with the task of adaptive FIR filtering with noisy input-output data. This method is similar to the BCRLS method which is proposed by authors recently in terms of use of introducing an auxiliary estimator but a different form with that one in BCRLS method. Several modified points both in theoretical discussion and recursive computing aspects in the new MBCRLS method lead to a reduction in computing cost and simple, readable and understandable derivation. Simulation results are conductive to verify the discussions.