Research And Implementation Of I/Q Calibration For Zero-IF Transceiver

The zero-IF transceiver has the characteristics of simple structure and easy integration.In recent years,it has been widely used in programmable RF transceivers.However,due to the imbalance of amplitude and phase in the I/Q branch during the circuit design process,the generation of the mirror image signal and the distortion of the signal constellation diagram bring difficulties to the subsequent signal processing.In this thesis,based on the digital domain,the I/Q calibration technology of the zero-IF transceiver is studied and implemented.After investigating the existing calibration technology,the three commonly used calibration ideas are analyzed and improved,and the corresponding design is proposed.The specific contribution of this thesis is as follows:Firstly,for the calibration technique based on the training sequence,the design of the joint calibration scheme for the transceiver based on the maximum likelihood estimation is completed.The estimation process is performed by combining the white noise of the transmitter,the receiver,and the channel.The entire estimation process requires only one set of training sequences,and can realize the separation of the unbalanced parameters at the transmitter and the unbalanced parameters at the receiver.The algorithm is verified by MATLAB simulation in the article.The simulation results show that the image rejection ratio at the transmitting end can reach 63 d B,and the image rejection ratio at the receiving end can reach 50 d B.Secondly,for the blind estimation calibration technology,the design of the real-time blind calibration scheme of the receiving channel based on the Fast Independent Component Analysis(Fast ICA)algorithm is proposed,and the hardware implementation and testing of the algorithm are performed.The algorithm does not require any priors about the signal Information and statistical characteristics,and the calibration process is not affected by the working state of the receiver,real-time calibration is achieved.In the hardware implementation process,all operations are completed using fixed-point numbers,saving power consumption and resource consumption.The test results show that the algorithm can improve the image rejection ratio of the tone signal to 69.3d B and the EVM of the modulated signal to-37 d B.Thirdly,for the calibration technology based on the statistical characteristics of the signal,the adaptive calibration technology of the receiving channel based on the variable step size LMS(VS-LMS)algorithm is completed by using the cyclic symmetry of the modulated signals such as M-PSK and M-QAM.After analyzing the existing variable step size algorithm,a new step size iteration formula is designed to improve the iteration speed and accuracy.The results of the test shows that the algorithm can improve the image rejection ratio of the single tone signal to 63.6d B and the EVM of the modulated signal to-35 d B,which reduces the resource consumption by nearly 50% compared with the second algorithm.The three calibration techniques designed in this thesis have different applications and their advantages and disadvantages.In particular,the latter two algorithms have completed FPGA-based testing,which lays the foundation for the on-chip implementation of the algorithm in the future.