A novel concept of embedding orthogonal basis function expansion in a feedforward neural equaliser

摘要

The proposed neural equaliser structure is based on an orthogonal basis function (OBF) expansion technique, motivated by genetic evolutionary concept, which utilizes a self-breeding approach to evolve new information so as to consolidate the final output.The equaliser structure developed using this novel approach has outperformed the conventional multilayer feedforward neural network (FNN) equaliser with a wide margin and its bit-error-rate performance is close to that of an optimal Bayesian equaliser. Also it learns faster with less training samples.Application of this proposed technique also reduces the structural complexity of a conventional FNN equaliser and has the potential to become a challenging candidate for real-time implementation issue.