Blind Frequency-Dependent I/Q Imbalance Compensation Using System Identification

A novel blind frequency-dependent I/Q imbalance compensation for direct-conversion receivers is proposed. By analyzing the filter which represents branch mismatches, we prove its minimum phase response, for practical front-ends with reasonable image attenuation quality. We also reveal that the received branch signals will have the same autocorrelation in the absence of frequency-dependent imbalance. The proposed method is to obtain the difference filter's magnitude response from the branch signals autocorrelations, and uniquely determine the corresponding minimum phase response afterwards. Then, the frequency-independent imbalance can be estimated subsequently from the branch signals cross-correlation. The proposed method has an analytical solution therefore can be easily implemented. Simulations confirm its superior performance in various I/Q imbalance scenarios.