Abstract: The Radical asis Function Neural Network(RBFNN) is a feed-forward NN model with the capacity of local approximation, so it can be used as a multiple-user detector(MUD) in the CDMA. Althouth it proves to be able to present a good performance in the detection, it has some disadvantages which lie in the determination of its structure and the optimization of its training algorithm. In this paper, a novel immune RBFNN MUD is proposed to balance computational complexity and performance of RBFNN. It is characteristic of more rapid convergence, better generalization and greater robustness, so its has better pracicability and real-time performance. Simulations also prove its effectiveness.

Key words: CDMA, MUD, RBFNN, immune