Portal de Periódicos da CAPES

A Kernel Affine Projection-Like Algorithm in Reproducing Kernel Hilbert Space

2019; Institute of Electrical and Electronics Engineers; Volume: 67; Issue: 10 Linguagem: Inglês

10.1109/tcsii.2019.2947317

1558-3791

Qishuai Wu, Yingsong Li, Yuriy Zakharov, Wei Xue, Wanlu Shi,

Blind Source Separation Techniques

A kernel affine projection-like algorithm (KAPLA) is proposed in reproducing kernel Hilbert space in non-Gaussian environments. The cost function for the developed algorithm is constructed by using the correntropy approach and Gaussian kernel to deal with nonlinear channel estimation. The devised algorithm can efficiently operate in the impulse noise. As a consequence, the proposed KAPLA algorithm provides good performance for nonlinear channel equalization in impluse-noise environments. Simulations results in different mixed noise environments verify the superior behavior of KAPLA compared to known algorithms.