Reduced-complexity genetic algorithm aided and radial basis function assisted multiuser detection for synchronous CDMA

Radial Basis Function Network aided Multiuser Detection (RBFN-MUD) schemes are capable of detecting the received signal of all users, even if the channel output states are linearly non-separable. However, their complexity may become excessive which renders their real implementation irrealistic, except when the number of users is low. In this contribution a novel reduced-complexity RBFN-MUD is developed, which invokes Genetic Algorithms (GAs) for reducing the number of RBFN-MUD centres. Our computer simulations showed that GAs are capable of considerably reducing the complexity imposed at the cost of a slight performance degradation.