Multiuser Detection Based On Intelligent Signal Processing

With the rapid development of intelligent signal process, the research of its application in telecommunication is widely carried out. Intelligent signal process, in combination with multi-user detection, can greatly suppress the interference. The main task of this paper is summarized as follows: ? A new method using Independent Component Analysis is proposed. The received energy of signals is controlled effectively, thus the near-far effect is overcome as much as possible; moreover, the effect that noise imposes on the detector is mostly reduced. Simulations verif~?the effectiveness of this method. ? On the basis of MMSE linear detector we use adaptive wavelet network to achieve minimum mean square error, and effectively eliminate multi-access interference. Simulations show the detector based on wavelet networks can converge rapidly and is near-far resistant. ? We employ GR.NN(Generalized Regression Neural Network) for demodulation of DS-CDMA signals in a multi-access environment. Compared with decorrelating detector and MMSE detector, this detector eliminate not only the need for on-line computation, but also the need to know the crosscorellations, or the signature waveforms of the desired user and the interfering users. On the basis of Adaptive MMSE linear detector we use GRNN to achieve minimum mean square error, and effectively eliminate multi-access interference. Simulations show the detector based on GRNN is near-far resistant.