Instantaneous Gradient Based Dual Mode Feed-Forward Neural Network Blind Equalization Algorithm

Ying Xiao

doi:10.19026/rjaset.5.5007

Abstract

To further improve the performance of feed-forward neural network blind equalization based on Constant Modulus Algorithm (CMA) cost function, an instantaneous gradient based dual mode between Modified Constant Modulus Algorithm (MCMA) and Decision Directed (DD) algorithm was proposed. The neural network weights change quantity of the adjacent iterative process is defined as instantaneous gradient. After the network converges, the weights of neural network to achieve a stable energy state and the instantaneous gradient would be zero. Therefore dual mode algorithm can be realized by criterion which set according to the instantaneous gradient. Computer simulation results show that the dual mode feed-forward neural network blind equalization algorithm proposed in this study improves the convergence rate and convergence precision effectively, at the same time, has good restart and tracking ability under channel burst interference condition.

Highlights

In the digital communication systems, finite bandwidth and multi-path propagation characteristics of the communication channels can cause severe InterSymbol Interference (ISI) in high data rate communication systems (Yin-Bing et al, 2010), which may lead to high error rates in symbol detection
We use the instantaneous gradient change rate to set threshold and a new dual mode neural network blind equalization algorithm was proposed, at last by using computer simulations proved the validity of the algorithm compared with decision circle based dual mode algorithm and sign error based dual mode algorithm, the performance for restart and tracking Ability under the channel has burst interference was done to prove the practical engineering value
The instantaneous gradient change rate reflects the network steady state and it can be used as the threshold for dual mode algorithm switching criterion

Summary

Introduction

In the digital communication systems, finite bandwidth and multi-path propagation characteristics of the communication channels can cause severe InterSymbol Interference (ISI) in high data rate communication systems (Yin-Bing et al, 2010), which may lead to high error rates in symbol detection. Blind equalization by feedforward neural network usually adopts the cost function of Constant Modulus Algorithm (CMA) (Yang et al, 2011), the convergence rate is very slow and the steady state error is big.

Results

Conclusion