Comparison of Nonlinear CMA-Criterion-Based Blind Equalizers

With a single-layer architecture and a linear transfer function (LTF), CMA cannot perform very well for the nonconvex and nonlinear cost function due to the convex decision region. Both nonlinear transfer function (NTF) and multilayer architecture (MLA) can generate a nonconvex decision region and thus improve the equalization performance. The objective of this paper is to investigate that it is NTF or MLA that does play an essential role in the performance improvement. To catch this goal, in this paper, three CMA-criterion-based methods including NTF-based CMA (NCMA), LTF-based multilayer CMA (MCMA) and NTF-based multilayer CMA (NMCMA) are proposed. Simulation of proposed algorithms on 16- QAM symbols indicates that the NTF is the prerequisite condition for the performance improvement, based on which MLA can futher enhance the equalization performance.