Adaptive equalization using complex-valued multilayered neural network based on the extended Kalman filter

In this paper, a complex-valued neural network based on the Kalman filter is presented for channel equalization in communication system. The complex-valued decoupled extended Kalman filter algorithm is derived which provides faster convergence rate and better performance than those of complex back-propagation algorithms. Computer simulation results showed that the proposed scheme has faster convergence behavior and smaller signal constellation than the gradient descent based learning algorithm via linear and nonlinear channel equalization problem.