Improved Optimum Error Nonlinearities Using Cramer–Rao Bound Estimation

In this paper, we propose an efficient design of optimum error nonlinearities (OENL) for adaptive filters which minimizes the steady-state excess mean square error and attains the limit mandated by the Cramer–Rao bound (CRB) of the underlying estimation process. Novelty of the work resides in the fact that the proposed improved optimum error nonlinearities (IOENL) design incorporates the effect of CRB which was ignored in the existing literature. To achieve this, we employ two efficient methods to estimate the variance of a priori estimation error. Therefore, the proposed IOENL does not use any assumption on the distribution of input regressor elements and noise sequence. Neither the assumption of independence on the input regressor is made nor any sort of linearization is assumed. Extensive simulations are done to show the efficiency of the proposed algorithm compared to the standard least mean square algorithm and the standard OENL algorithm.