Generalized derivation of neural network constant modulus algorithm for blind equalization

Authors:
Donglin Wang
Affiliations:
Department of Electrical and Computer Engineering, University of Calgary, Calgary, Alberta, Canada
Venue:
WiCOM'09 Proceedings of the 5th International Conference on Wireless communications, networking and mobile computing
Year:
2009

Citing 2
Cited 0

Feedforward neural network for blind equalization with PSK signals

Neural Computing and Applications
Nonlinear blind equalization schemes using complex-valued multilayer feedforward neural networks

IEEE Transactions on Neural Networks

Quantified Score

Hi-index	0.00

Visualization

Abstract

With the single-layer architecture and linear transfer function, constant modulus algorithm (CMA) cannot work well for the nonconvex and nonlinear cost function due to the convex decision region. To suppress the convergence error and improve the performance, the neural network constant modulus algorithm (NNCMA) is proposed by integrating CMA and neural network scheme. However, all existing NNCMAs are not more than two layers, which obstructs the futher study on NNCMA. In this paper, the generalized derivation of NNCMA for any given layer is thoroughly developed by using backpropagation algorithm. Simulation of NNCMA on 32-QAM symbols indicates a much better equalization performance is achieved relative to CMA.