Reduced complexity in-phase/quadrature-phase turbo equalisation using radial basis functions

摘要

A novel reduced complexity Radial Basis Function (RBF) neural network based equalizer, referred to as the In-phase/Quadrature-phase RBF Equalizer (I/Q-RBFEQ), is proposed. The I/Q-RBF-EQ is employed in the context of turbo equalization (TEQ) assisted by iterative channel estimation. The performance of the I/Q-RBF-TEQ is characterized in a noise limited environment over an equally weighted, symbol-spaced three-path Rayleigh fading channel. The I/Q-RBFTEQ achieved the same performance as the conventional turbo equalizer, while achieving a complexity reduction by a factor of 1.5 and 109.6 for 4-QAM and 16-QAM, respectively.