This thesis investigates digital signal processing (DSP) techniques for compensating both the I/Q mismatch and the DC offset in communication receivers.
First, a data-aided adaptive I/Q mismatch and DC offset compensation method is developed in frequency-selective channel environments. This method is derived by extending the existing compensation technique, which is derived in AWGN environment, to frequency-selective channels. The method compensates for not only the I/Q mismatch and the DC offset but also the inter-symbol-interference (ISI). The optimal filter tap coefficients are analyzed in a mean-square error (MSE) criterion. After evaluating theoretical performance of the compensation method, a filter tap coefficient initialization scheme is developed using the recursive least-squares (RLS) algorithm.
In the second part, a new I/Q mismatch and DC offset compensation method is proposed. In this method, the I/Q mismatch and the DC offset are estimated using the least-squares (LS) method with the aid of a training sequence. The I/Q mismatch and the DC offset are eliminated using the estimates. In addition, a group of training sequences that minimizes the MSE of the estimate is determined. Since this compensation scheme is based on an estimation technique, it requires a shorter training sequence than the above compensation method which is based on an adaptive filtering technique. The advantages of proposed method are demonstrated through computer simulations.
Finally, the proposed I/Q mismatch and DC offset estimation method is extended to time-varying channel environments. To extend the estimation method to fast fading channels, polynomial time-varying channel model is employed. Through computer simulations, it is shown that the estimation method exhibits robust performance in fast fading channel environments.