Maximum Correntropy Criterion-Newton Algorithm for Adaptive Interpolated Volterra Filter

The main drawback of the second-order Volterra (SOV) filter is that its coefficients increase exponentially with the length of the memory, which has promoted the development of sparse-interpolated Volterra filters. However, the existing algorithms only investigated the performance of least-mean-square (LMS)-based interpolators, and their performance may degrade for combating impulsive noise. To address this problem, the maximum correntropy criterion-Newton algorithm for adaptive interpolated Volterra (MCCN-AIV) filter is proposed, leveraging faster convergence of the Newton method and numerical robustness of the MCC. Additionally, the MCCN-AIV algorithm is analyzed by utilizing the energy conservation argument (ECA) and the Taylor expansion method. The accuracy of analysis and efficacy of the MCCN-AIV algorithm are validated by nonlinear system identification, nonlinear acoustic echo cancellation (NLAEC), and chaotic time-series prediction.