Fixed-point maximum complex correntropy algorithm with variable shape parameter

As a similarity measure for complex random variables, the complex correntropy algorithms have received much attention for their robustness in the face of non-Gaussian disturbances. However, since the existing relevant studies mainly focus on the complex correntropy of Gaussian kernels with a single kernel, this greatly limits the flexibility of correntropy. In order to improve the flexibility of correntropy, this paper proposes a fixed-point maximum complex correntropy adaptive filtering algorithm with variable shape parameter (VSPMCCC-FP) in the kernel function, in which the shape parameter varies with the change of the real part of the system error. Then we analyze the stability of the proposed algorithm. Finally, we test the performance of the proposed algorithm in a system identification scenario, and the results show that the VSPMCCC-FP algorithm has faster convergence performance and lower steady-state misalignment than other competing algorithms.