Fast Bias Removal Equation-Error Adaptive Filtering Algorithms.docVIP

IAENG International Journal of Computer Science, 35:1, IJCS_35_1_19 ______________________________________________________________________________________ Fast Bias Removal Equation-Error Adaptive Filtering Algorithms Junghsi Lee, Yi-Wen Chiu, and Hsu Chang Huang multimodal and therefore the output-error algorithm may Abstract—This paper presents a multi step-size monic converge to the local minimum. normalization equation-error linear filter. We then extend the idea to nonlinear adaptive filter and derive a multi step-size monic normalization equation-error bilinear filter (MSS MNEEBF). The algorithms enjoy fast convergence behavior and can remove biased estimates associated with conventional equation-error adaptive filters. Simulation results validate the usefulness of our algorithms. The equation-error approach is very simple to design and implement. It does not need stability check. The error surface is unimodal. However, it results in biased system estimates in the presence of noisy observations. Consequently, such equation-error algorithms are useful only in applications in which the measurement noise is fairly small. Early equation-error IIR algorithms tried to remove the bias Index Terms—Bias-Removal Algorithms, Bilinear Adaptive through additional filtering or noise suppression schemes [1]. Filters , Equation-Error Adaptive Filters, Nonlinear Adaptive Filters Some alternatives such as the unit-norm constraint approach [2], [3] and monic normalization method [4] have been proposed to produce unbiased parameter estimates in the presence of noise. The basic idea behind the monic normalization equation-error I. INTRODUCTION filter (MNEE) is to set the first denominator coefficient of a Finite impulse response (FIR) least mean square (LMS) rational system to unity, which is implemented by having all its adaptive filter h