Estimation and compensation of I/Q mismatch and frequency offset for direct-conversion and low-IF receivers

Abstract

This thesis investigates a set of new baseband digital signal processing (DSP) methods for estimating frequency offset and I/Q mismatch in a frequency selective channel for direct-conversion and low-IF receivers. First, a technique (MLE#1) is proposed that jointly estimates the I/Q mismatch, DC offset, carrier frequency offset, and channel for direct-conversion receivers in frequency selective channels. The joint estimator is a data-aided technique that requires a training sequence and is derived based on the maximum likelihood (ML) criterion. The characteristics of the joint estimator are analyzed. In particular, it is shown that its accuracy almost achieves the $Cramér-Rao$ lower bound (CRLB). In addition, the issue of designing an optimal sequence for the proposed estimator was addressed by developing a measure to qualify a sequence. The advantage of the proposed estimator over an existing method and the validity of the designing method of an optimal sequence are demonstrated through computer simulation. In the second part, a joint frequency offset and I/Q mismatch estimator is developed for OFDM systems by exploiting the repeated preamble structure. The proposed estimator is featured by its reduced complexity and better performance compared to the estimator MLE#1 adapted to OFDM signals. The performance of the proposed method is evaluated through simulation. Finally, a technique is presented that can compensate for the I/Q mismatch in low-IF receivers. In particular, a non-data-aided (NDA) I/Q mismatch estimator is derived by exploiting the statistical independence between the desired and image signals. The mean square error (MSE) of the estimate is analyzed. Computer simulation results indicate that the proposed technique can outperform existing adaptive DSP techniques that are based on the use of blind signal separation algorithms. It is observed that the image rejection ratio (IRR) of the proposed technique decreases monotonically with the numb...