The Research On Blind Beamforming Algorithm Based On Approximation Of Function By RBF Network

During the past decade, wireless communication systems have made a great progress world-wide. As the key technologies, the application of smart antenna technique in wireless communications allows the base station to transmit to and receive from mobiles by directional beams, which can improve the system performance by relieving the CCI, decreasing the transmission power,improving the spectrum efficiency,enlarging the system capacity and so on . Smart antenna has become a very promising technology for wireless communication,and beanforming is the core of the smart antenna.In the 3G mobile communication system,Multiple Access Interference and ISI are the two main sources of interference.So presenting a high performance. low complexity algorithm is the effective way to slove this problem. This paper introduces a scheme of embedding the blind beamforming into the neural network and rake receiver. The main works in this paper can be summarized as follows:First: At the beginning part of this thesis, the principle,structure,work pattern and expression of smart antenna are introduced conceptually, the several vector optimization rule are discussed and compared ,the beamforming methods for smart antenna are explored ,the blind adaptive beamforming algorithms are introduce emphatically. we establish the CDMA signal model of sigle-path and multi-path environment. The algorithm based on code filtering and improved algorithm are deduced in detail.Second: We deeply study the blind beamforming algorithms which adaptive to DS-CDMA system.Due to the limitations of these algorithms ,a approach of the beamforming with radial basis function neural network for smart antenna in the mobile communication systems is suggested.Accoding to the characteristic of CDMA system on the Gauss White Noise signal channel,steering vector is estimated by higher-order cumulant,and weight vector is approximated by RBFNN to carry out the optimum blind bemaforming.Theoretical analysis and MATLAB simulation prove this method can not only make the computation rapid but also is robust to system error.Third: The channel is multi-path slow fading effect in the mobile communication systems, which show time-delay-dispersion and space-angle-dispersion.Thank to this interference , a adaptive algorithm the parallel interconnection of smart antenna and Rake reeiver is presented. This algorithm can catch each of the multi-path components of the interested user in 2-D space-time domain, and combine these components coherently and simultaneously,while suppress multiple-access interference(MAI) and inter-symbol interference(ISI) effectively.The move ability will improved more when composed with the neural network.Trough simulation,the space-time adaptive blind beamforming algorithm based on ANN is application greatly.The research shows that the scheme which the space-time connection adaptive algorithm based on the neural network can improve the performance and the low complexity of algorithm, and avoid MAI and ISI basically.