Semi-Blind structured subspace method for signal estimation in nonlinear convoluted mixture

Abstract

This work proposes a novel semi-blind structured subspace signal estimation approach for nonlinear single-input multiple-output (SIMO) systems. Here, in this work, a quadratic nonlinearity is considered in the analysis. The proposed approach exploits the Toeplitz structure present in a convolutive communication model together with pilot signals in a semi-blind scheme. It directly estimates the signal with no channel estimation step, which saves the extra computational cost needed when considering a two steps approach (i.e., channel identification then equalization). The simulation findings show that the proposed method is resilient to channel order overestimation errors and allows achieving good signal restoral quality with only few pilot symbols, outperforming the pilot-based approach in such a context.