The Research Of Blind Multiuser Detection Using Neural Network Based On Constraint Constant Modulus Algorithm

With the application of the third generation mobile system, CDMA as the main access method is developing quickly these years. However, multiple access interference and near-far effect which CDMA possesses restrict system capability, regular filter can not overcome these questions. As one of the most important technologies of CDMA, Multiuser Detection (MUD) technique has a good performance of resisting MAI and near-far effect through the use of adequate cross correlations among spreading codes of users to detect the signals. So blind multiuser detection has attracted considerable attention and significant researches. Recently, blind multiuser detection based on neural network has become a research focus, integrates merits needing a few using informations of blind MUD with merits of neural network such as quickly speed, strong processing, and so on.The major works of this paper are summarized as follow: 1. Briefly looks back the development of the third generation mobile communication system and the key technologies of 3G.On the foundation of discussion, we explain the necessary to research of the Multiuser detection. The principle of MUD is introduced in this paper, and the characteristics, performance and drawbacks of the existed neural network algorithms is discussed.2. A lot of current algorithms based on Hopfield neural network are analysed. Pointing to the drawback, two new algorithm of Blind MUDbased on constraint CMA for realized DS-CDMA have proposed by using Hopfield neural network through the merge of penalty function and energy function of Hopfield. Numerical results demonstrate the two proposed approaches are more superior to present algorithms.3. Aiming at the drawbacks of the existed CMA algorithms, we introduce update constraint CMA. On the foundation of the new Constant Module Algorithm, two improved Hopfield neural network has been proposed. The simulation results show the two improved algorithms is more superior in computational efficiency and real time operation to traditional algorithms.