Nonlinear equalization with known channel state information in satellite communication

Three nonlinear equalization algorithms are discussed for the case that training sequence is not available but channel state information (CSI) is known. The first algorithm is based on the extended Kalman filter (EKF). The second one is based on decision feedback equalization (DFE), where a modified DFE (MDFE) is proposed. Both of the two algorithms use CSI directly. Independent of CSI, the third equalization algorithm is based on complex bilinear recurrent neural network (CBLRNN) and stop-and-go (S&G) algorithm. Simulation results show that MDFE is much better than the other two algorithms for nonlinear channels with either severe or mild intersymbol interference (ISI).