A kernel canonical correlation analysis algorithm for blind equalization of oversampled Wiener systems

In this paper we present an algorithm for blind equalization of single-input multiple-output (SIMO) nonlinear systems, in which each nonlinear channel is a Wiener system. The proposed method combines ideas from blind linear SIMO identification with kernel canonical correlation analysis (kernel CCA) to identify the nonlinearities. It is shown in the paper that the blind equalization problem can be solved in an iterative manner, alternating between a CCA problem (to estimate the linear filters) and a kernel CCA problem (to estimate the memoryless nonlinearities). The resulting algorithm can be applied to the general case of nonlinear SIMO systems with P outputs. Simulations are included to demonstrate its effectiveness.