“…Although linear equalization is the simplest equalization method, it delivers an extremely inferior performance compared to that of the optimal methods, such as Maximum A Posteriori (MAP) or Maximum Likelihood (ML) methods [10,17,18]. Nonetheless, the high complexities of the optimal methods, and also their need of the channel information [8,10,12,19,20], make them practically infeasible for UWA channel equalization, because of the extremely large delay spread of UWA channels [8,18,[21][22][23]. Hence, we seek to provide powerful nonlinear equalizers with low complexities as well as linear ones.…”