Blind Equalization in Neural Networks

3Research of Blind Equalization Algorithms Based on FFNN

Abstract: In this chapter, the basic principle of feed-forward neural network (FFNN) is analyzed. First, blind equalization algorithms based on the three-layer FFNN, four-layer FFNN, and five-layer FFNN are studied. Then iteration formulas of algorithms are derived. Computer simulations are done. The theoretical analysis and experimental results verify that with the increase of layer number, the algorithm convergence rate becomes slow and the computational complexity increases. But the steady residual error decreases after the algorithm converged, that is, the approximation ability enhances. Second, the improved BP algorithm is applied to the blind equalization algorithm, then blind equalization ...