Modified Error Data Normalized Step Size algorithm Applied to Adaptive Noise Canceller

Abstract: This paper introduces a Modified Error Data Normalized Step Size (MEDNSS) algorithm in which time varying step size depends upon normalization of both error and data vector. An Adaptive Noise Canceller (ANC) is used to improve the system performance in the presence of signal leakage components or signal crosstalk. This ANC consists of three microphones and two adaptive filters that automatically adjust their filter coefficients by using MEDNSS algorithm. The first adaptive filter cancels the signal leakage components and second adaptive filter cancels the noise. The performance of the MEDNSS algorithm is analyzed, simulated and compared to the Error Data Normalized Step Size (EDNSS) algorithm in stationary and non-stationary environments using different noise power levels. Computer simulation results demonstrate the significant improvements of the MEDNSS algorithm over the EDNSS algorithm in minimizing the signal distortion, Excess Mean Square Error (EMSE) and low misadjustment factor