DSP implementation of a neural network based blind multiuser receiver for DS-CDMA communication systems

Broadband wireless communications are going to be deployed widely due to the introduction of the Universal Mobile Telecommunications System (IMT/2000), which is based on wideband-code division multiple access (W-CDMA) techniques. Adaptive blind multiuser detection, proposed for the mobile station receiver, reaches the optimum solution after a certain number of bit times. We propose the implementation on a fixed point digital signal processor (DSP) architecture of the adaptive blind algorithm based on a neural approach. A modified Kennedy-Chua (1988) neural network, based on the Hopfield model, is proposed, and its implementation on a fixed point numerical environment has been studied by means of computer simulation. Finally we discuss a medium level implementation of the algorithm on a Texas Instruments DSP.