Implementation of Normalized Data Non-Linearity and Constraint Stability Least Mean Square algorithm with a new SuFee model for Adaptive Noise Cancellation

This paper discusses the implementation of variants of LMS (Least Mean Square) algorithm on new model named as Suspended Feedback (SuFee). Implementation of LMS (Least Mean Square) algorithm along with its variants like Constraint Stability Least Mean Square (CSLMS) and Non-Linear Data Least Mean Square (NDLMS) with conventional model and new SuFee model has been done. This model is an advanced version of conventional adaptive noise canceller. It is a simulation-based signal flow model. The conventional and SuFee models are evaluated and simulated using MATLAB, taking speech as an input signal mixed with white noise. The new SuFee model provides an advantage of faster error convergence with significant improvement in the output quality of the signal. Convergence was evaluated by displaying the learning curves and error signals of the different adaptive filter algorithms. Signal to Noise Ratio(SER) is the parameter to evaluate the quality of the output signal. The result shows the signal quality has been enhanced by three times better than the conventional model.