Conclusions

We presented a novel K-CCA algorithm for the supervised equalization of nonlinear Wiener systems, exploiting the system structure. We also developed an online version of this algorithm, which combines a sliding-window approach with a reformulation of CCA as an iterative regression problem. Simulation examples show fast equalization of time-varying Wiener systems and results of the influence of the different algorithm parameters on its performance were presented. In particular, if the length of the Wiener system filter is not known and overestimated, the algorithm performance is hardly affected.