New iterative kernel algorithms for nonlinear acoustic echo cancellation

Abstract

Recently, a nonlinear acoustic echo cancellation algorithm based on the framework of kernel methods has been proposed by modeling the echo path as a Hammerstein system. However, it requires a large amount of computation to be implemented. In this paper, we propose to use iterative methods for solving linear systems in order to reduce the numerical complexity of identifying the nonlinear and linear parts of the echo path. Also we investigate the effect on performance of the parameters of the methods for both iterative batch and online versions. Simulation results confirm the complexity reduction and good performance of the proposed methods for clipping nonlinearity and room impulse response variation.

Hierarchical gradient- and least squares-based iterative algorithms for input nonlinear output-error systems using the key term separation

“…Therefore, we can get the key term separation identification model in (20) of the IN-OE system in ( 2)-( 1), where ϑ is the parameter vector to be identified and contains all parameters of the original system. Remark 2: Obviously, the dimension of the parameter vector ϑ of the key term separation identification model in (20) is smaller than that of the over-parameterization identification model in (12).…”

“…Various parameter identification methods have been proposed for input nonlinear systems, such as the auxiliary model identification idea [18], the multi-innovation identification theory [19]. For example, Albu and Nishikawa presented an iterative kernel algorithm for nonlinear acoustic echo cancellation [20]. Equation-error models and output-error models are two basic types of stochastic systems, and have received considerable attention in the field of system identification [21].…”

Filtered multi‐innovation‐based iterative identification methods for multivariate equation‐error ARMA systems

“…10 Albu and Nishikawa proposed a new iterative kernel algorithm to deal with the nonlinear acoustic echo cancellation. 11 For a class of nonlinear least squares problems, Gan et al introduced a variable projection algorithm by separating the variables into a linear and a nonlinear part and taking full advantage of liner least squares techniques. 12 After representing the sinusoidal signal as a linearly parameterized form, Na et al developed an adaptive identification framework for sinusoidal signals to estimate the unknown amplitude, frequency and phase, where only the output measurements are used.…”

“…Pu and Bai proposed a robust recursive least squares identification method for the discrete‐time linear systems with unknown time‐varying disturbance 10 . Albu and Nishikawa proposed a new iterative kernel algorithm to deal with the nonlinear acoustic echo cancellation 11 . For a class of nonlinear least squares problems, Gan et al introduced a variable projection algorithm by separating the variables into a linear and a nonlinear part and taking full advantage of liner least squares techniques 12 .…”

Diffusion Hammerstein Spline Adaptive Filtering Based on Orthogonal Gradient Adaptive Algorithm

“…Hammerstein architecture is the block-oriented nonlinear model consisting of a cascade topology with memoryless functions in the linear time invariant (LTI) model called as nonlinear-linear model [13]. Hammerstein system has been used as a kernel method modelled for the echo path in a nonlinear acoustic echo cancellation [14], [15]. Hammerstein VOLUME 4, 2016 SAF (HSAF) has been proposed on the stochastic gradient scheme in the impulsive noise environment [16]- [18].…”