BLIND EQUALIZATION USING HIGHER-ORDER CUMULANTS AND NEURAL-NETWORK

This paper develops a new method to achieve blind equalization in digital communication for linear finite impulse response (FIR) systems, whether the systems are minimum phase or not. This new approach divides the problem into two parts. First, it employs the characteristic of the linear system to estimate the original channel based on the fourth-order cumulants instead of time samples of the channel output. Thus, nonminimum phase channels can be handled. Second, it utilizes the nonlinear characteristics of the neural network to build an inverse system (equalizer) for the original channel. This is done by using the estimated channel as a reference system to train the neural network. The neural network helps the equalizer to reduce the degree of model uncertainty and makes the equalizer resistant to additive noise. Taking the advantages of both linear and nonlinear systems, this new scheme works well for both stationary and nonstationary cases and leads to good equalization results.