Nonlinear Bitwise Equalization
A new family of nonlinear equalizers that detect Quadrature Amplitude Modulated (QAM) signals by decomposing the process into a binary decision “cascade” is proposed. First, the quadrant of the symbol is determined; second, we determine which quarter of the chosen quadrant contains our symbol, and so on. We have used two nonlinear architectures to approximate the optimal decision boundaries: a Volterra filter and a Generalized Cerebellar Model Arithmetic Computer. Our analysis demonstrates that bitwise equalizers achieve lower Bit Error Rate (BER) than transversal ones.
KeywordsFinite Impulse Response Quadrature Amplitude Modulate Cerebellar Model Articulation Controller Committee Machine Volterra Filter
Unable to display preview. Download preview PDF.
- 2.Figueiras-Vidal, A.R., Martínez-Ramón, M., Artés-Rodríguez, A., Cid Sueiro, J.: Iterative Decision for Multilevel Equalization. In: Proceedings of the IASTED Int. Conference Expert Systems and Neural Networks, Hawaii, USA, pp. 357–360 (1996)Google Scholar
- 3.González-Serrano, F.J., Artés-Rodríguez, A., Figueiras-Vidal, A.R.: Generalizing CMAC Architecture and Training. IEEE Transactions on Neural Networks 9(6) (1998)Google Scholar