A blind-ML Method for frequency-selective I/Q mismatch compensation in low-IF receivers

This paper presents a blind method to resolve imbalances between I and Q paths in low-IF receivers. In particular, we investigate the mismatch arising from the coefficient variation of the complex filters that leads to the frequency-selective imbalance problem. The key point in the proposed novel DSP-based solution is the transformation of the signals to frequency domain and the estimation of the mismatch cancellation filter based on the instantaneous separation in each frequency bin. Although this approach relaxes the original model with convolutive mismatch, it still requires long observation records to obtain a sufficient performance. The proposed method exploits the continuity property of the cancellation filter in the frequency-domain to improve the estimator performance even with short symbol block lengths. Moreover, we derive an asymptotic bound on the attainable MSE in the case of Gaussian sources and deterministic parametrization of the frequency-selective model. The numerical results under the mismatched filtering condition reveal that the interference-separation performance of the proposed method is superior to that of the conventional methods and approach the performance bound for interference-to-signal ratios of interest.