| With the rapid development of electronic information technology,the complexity and transient characteristics of electronic signals are showing a sharp upward trend,which puts forward higher requirements for system bandwidth in the fields of wireless communication,electronic measurement,digital medical treatment,and radar testing.The signal to be analyzed in a broadband system is often located in the radio frequency(RF)frequency band,the direct sampling of the RF signal has many technical limitations and hardware defects with the relatively high cost.The use of analog mixing technology to down-convert the RF signal before digital sampling is an effective solution to the above problems.The zero-IF and low-IF structures can greatly alleviate the requirement of the sampling rate and analog input bandwidth of analog to digital converters(ADCs)due to the use of the quadrature mixing structure,which have been widely used in recent years due to their low cost and high efficiency.The quadrature mixing structure uses Hermitian symmetry of the real signal spectrum to eliminate the image frequency,so there is no need for an image rejection filter,which makes the structure more suitable for integration.However,the analog components that the structure uses have non-idealities inevitably,introducing the inconsistent frequency response characteristics of the inphase/quadrature(I/Q)channels and causing gain and phase mismatch of the I/Q signal,which is called the I/Q imbalance.I/Q imbalance leads to insufficient rejection of the image frequency,seriously deteriorating the relevant indicators of the system,affecting system performance and correct analysis of the back-end signal.In a wideband system,the imbalance problem will be more serious than that of the narrowband and become one of the leading reasons that affect system performance,and the main bottlenecks for designing high-performance systems at the same time.Based on the above-mentioned background and faced problems,the main work of this dissertation is as follows:(1)A detailed analysis of the limitations of the existing I/Q imbalance model as for broadband systems is presented,and the serious impact of time mismatch(TM)error on the broadband system is fully studied.Based on the above research,an augmented error model of wideband I/Q imbalance including TM errors is established,and a research route based on the augmented error model to in-depth study the broadband I/Q imbalance problem is established.Based on the different representations of various errors in the augmented error model,the augmented error model is further refined into static imbalance error model and dynamic imbalance error model.(2)A static phase error estimation algorithm based on the cross power spectrum between I/Q signals is proposed,combining the unwrapping method based on the extended Euclid algorithm(EEA)to eliminate the influence of phase unwrapping to complete the static phase error estimation.The static amplitude error is obtained by dividing the amplitude spectrum of the corresponding frequency point of the I/Q signal.Then a compensation structure was designed according to the estimated static imbalance error.The fractional delay(FD)filter designed by the window function method was used to compensate the fractional TM error,and the design results of different window functions are compared through simulation experiments,and the window function type with the best design result is selected.The FD filter is converted into a Farrow structure with variable time delay by polynomial fitting,which improves the flexibility of the structure and the compensation efficiency,and provides a foundation for the compensation of static imbalance errors.(3)Based on the related research of the preamble static error,the estimation and compensation strategy of the dynamic imbalance error is designed.Taking into account the characteristics of dynamic error parameters changing with frequency,two error estimation methods based on sine fitting estimation based on time domain characteristics and frequency domain estimation based on spectrum measurement are analyzed in detail,and then the idea of divide and conquer method is used to compensate the dynamic amplitude and phase error.The dynamic amplitude error compensation adopts a finite impulse response(FIR)filter with linear phase(LP)characteristics,which is designed by the frequency sampling method.Based on the nonlinear phase(NP)error in the dynamic error model,an all pass filter(APF)is added to the static phase error compensation and expands the original compensation structure into a dynamic phase error.The meta-heuristic algorithm,flower pollination algorithm is used to optimize the design of the phase compensation structure.In order to ensure the stability of the design APF,the design problem of the compensation structure is transformed into a nonlinear optimization problem with constraints to obtain the optimal compensation structure parameters.(4)On the basis of the research of divide and conquer method,a joint compensation strategy of NP error and amplitude error in the dynamic imbalance error model is further proposed.The estimated phase error is decomposed by a polynomial fitting method,and on the basis of the original static error compensation framework,a nonlinear phase FIR filter is used to jointly compensate the amplitude error and the NP error obtained by polynomial fitting.The design problem of the nonlinear phase FIR filter is transformed into the problem of solving the linear system,which is solved by the conjugate gradient squared(CGS)algorithm of the Krylov subspace iteration method.In order to ensure the design accuracy of the filter passband and the smooth transition band attenuation characteristics,the CGS-based filter design further incorporates the design idea of increasing the transition band frequency interval to reduce the number of transition band design points.The design idea is proposed to effectively improve the relative image ratio(RIR)and avoid the deterioration of the system signal-to-noise ratio.Using the idea of "offline estimation,online compensation",this dissertation proposes different estimation methods and compensation frameworks without adding any additional hardware circuits for different imbalance models combining data-assisted methods and time-domain filtering structures,which has strong implementability.Based on the above research content,the key algorithm proposed in this dissertation are realized on the platform of the broadband vector signal transceiver project developed by my team with a sampling rate of 1.25 GSPS and a bandwidth of 1 GHz.The finite word length effect of the filter realization in field programmable gate array(FPGA)is analyzed,and the parallel filter structure is adopted to shorten the filtering processing time and enhance the real-time performance of the compensation structure.The average system RIR are largely improved from 4.58 dB to 68.31 dB after compensation by the designed structure,which has verified the effectiveness of the design algorithm for I/Q imbalance in the real broadband system,which guarantees the subsequent system to detect and analyze the signal correctly. |
PDF Full Text Request