Channel equalization is essential process in mobile communication to compensate the distortion of received signal resulted from channel effect. In this paper, multi-layer perceptron with decision feedback is used for nonlinear satellite channel equalization, and the proposed structure shows the BER performance similar to MLSE(Maximum Likelihood Sequence Estimator).
For VLSI implementation, a neuron cell and a synapse cell are designed in schematic level. and simulated with Hspice. To evaluate learning capability, XOR problem are tested. Finally, with computer generated data, channel equalization with designed circuit are simulated.