A Modified Least Mean Square Algorithm Using Fractional Derivative and its Application to System Identification

Date Added: Aug 2009
Format: PDF

In this paper, a modified Least Mean Square algorithm has been developed by incorporating the fractional derivative term as well in the weight adaptation equation of standard Least Mean Square (LMS) algorithm. Performance analysis of the algorithm was carried out by applying it for a typical system identification problem. The results obtained using the newly developed algorithm, were compared with the standard LMS algorithm for different step size parameters and different number of tap weights. It was observed that the designed algorithm has faster convergence than the conventional LMS algorithm. It takes almost three times less number of iterations as compared to the LMS algorithm.