The Research Of Signal Processing Technology With Multiple Antennas In Cognitive Radio

In the information society,with the continuous improvement of people's living standards,people's demand for wireless communications is constantly expanding.However,the wireless spectrum resources are not infinite,which leads to the contradiction between the limited wireless spectrum resources and the growing demand for communication.As a kind of intelligent communication system,cognitive radio technology can improve the utilization of spectrum resources.Signal processing technology is the basis and key to the operation of cognitive radio system.In different stages of cognitive radio communication,the focus of signal processing technology is also different,such as in the spectrum sensing phase,signal processing is aimed to detect the primary user signal,while in the spectrum decision phase,it is aimed to estimate the channel information on the spectrum hole.Therefore,the further research and development of signal processing technology will make the cognitive radio technology get better development and application.This dissertation does research on the signal processing technology applied in the cognitive radio of signal' detection,the DOA estimation and signal separation.The main contents and results are summarized as follows:1.The spectrum sencing with muiltiple antennas is studied.Through the analysis of the classical energy detection method,it can be known that the detection performance of the energy detection method will be affected by noise uncertainty.In order to reduce the influence of noise uncertainty,a method of multiple antennas signal detection based on energy detection is proposed.The improved detection method is similar to the energy detection method,and the final decision is constructed by simply combining the signals obtained from the multiple antennas.Through the theoretical analysis of the newly constructed decision,it indicates that the calculation of the decision threshold is independent of the noise variance,and the detection probability expression is also a function independent of the noise uncertainty.Therefore,the proposed method can detect the primary user without noise variance estimation,while the energy detection method will lead to the detection performance loss due to the noise uncertainty.Because of the simple combination,the method is easy to implement like the energy detection method.The simulation results show that the improved detection method with three or more anntennas can obtain better detection performance than energy detection method,and also will not be affected by the impact of inaccurate estimation of noise variance.2.For DOA estimation,three DOA estimation methods(PM?DOA?MEMP)are analyzed.First,an automatic pairing DOA estimation method based on PM is proposed.In this method,a parallel uniform linear array structure with position offset is designed,which can be used to obtain a diagonal matrix with paired information.By combining the output of the array with certain rules,the estimation of the diagonal matrix can be obtained by using the PM algorithm and matrix decomposition,thus the estimation of DOA is achieved.Then,the DOA matrix method is analyzed and improved.In the parallel array structure,the estimation of DOA can be directly obtained through the eigenvalue matrix by using the position offset of structure;and in the symmetric V shape array structure,the DOA matrix method is improved and expanded by using rotation invariance of the sub-array and transformation calculation of the direction matrix.In the end,an improved MEMP method is proposed for a special dual parallel array structure.The method can work under the condition which is that the antennas number in a sub-array is smaller than the source signal number,and all the DOA of source signals can be estimated and automatic paired by structuring the Hankel data matrix and using the position offset of array structure.Simulation results show that the proposed three methods can achieve a certain accuracy estimation of DOA,while the additional pairing computation is not needed.3.The signnal sepration with uniform linear arrays is studied.The mixed matrix of the signal has a special characteristic because of the structure of uniform linear arrays,and the linearization of the output signal is multiplied by the linear transformation of the special mixed matrix.At the same time,the array output can be used to obtain the separation matrix which is corresponding to the mixed matrix directly by using the matrix decomposition and the ESPRIT algorithm.Without the linearization of the output signal,the source signal aslo can be obtained by using PM algorithm.The simulation results show that the signal separation based on DOA estimation is feasible under the condition of no noise.