Researeh On Adaptive Beamforming Algorithm Of Smart Antenna

As the number of mobile users and data increase sharply, the capacity demand for next generation communication. As one of the key techniques for next generation, the smart antenna has played an increasingly significant role.The adaptive beamforming algorithm is the technique of smart antenna. In this thesis, it mainly studies adaptive beamforming algorithms of the smart antenna technology. The thesis introduces research background, trends of the smart antenna and its advantages. Then, it has sketches the basic operation principle, classification and signal model of the smart antenna.Next, the thesis introduces beamforming theory and the definition of the antenna pattern. The relevant parameters such as the weighted vectors, the different element spacing, the number of array elements in antenna pattern are discussed.Subsequently, detailed description and derivation of beamforming filtering criteria are finished in this thesis. For example, minimum mean square error (MMSE) criterion, least squares (LS) criterion, minimum variance (MV) criterion and the maximum SINR (Max-SINR) criterion. In the end of this part, we made a comparison from the aspect of cost function, the optimal weights and the advantages and disadvantages of the criterion.Based on the best adaptive beamforming criterion, the thesis analyzes and contrasts the several non-blind adaptive beamforming algorithms. For the traditional Least Mean-Square (LMS) algorithm which is affected by the step size in convergence speed, tracking abilities of time-varying system and steady state imbalance between contradictions, the thesis has given three kinds of new variable step size for LMS algorithm:Normalized LMS, LMS based on Sine Function and Sigmoid Function. The MATLAB simulation results prove that the proposed algorithm has superior performance on convergence speed, tracking abilities of time-varying system and steady state imbalance between contradictions.Finally, the thesis has introduced Constant Module Algorithm (CMA) of blind beamforming algorithm in briefly. For Stochastic Gradient CMA(SGD-CMA) algorithm which is affected by the step size in convergence speed, tracking abilities of time-varying system and steady state imbalance between contradictions, the thesis propose a new variable step size CMA algorithm: Normalized SGD-CMA.The MATLAB simulation results prove that the new variable step size algorithm has solved the contradiction between convergence speed and steady state for the SGD-CMA algorithm, and the convergence performance of the algorithm is improved. At the same time, it also forms larger power gain in the direction of the desired signal, the "nulling" in the direction of the interference signal, and effectively suppresses the interference.