Non-Data-Aided I/Q Imbalance Compensation Using Measured Receiver Front-End Signals

The direct-conversion or homodyne receiver principle has received lots of interest in recent years for building flexible and compact radios. The homodyne architecture is, however, found to be rather sensitive to certain non-idealities of the analog front-end components. One big concern here is the amplitude and phase mismatches of the I and Q signal branches, resulting in insufficient rejection of the mirror frequency band. This paper presents an efficient low-order statistics based method (stat) for digital I/Q mismatch compensation, and compares the proposed technique with the blind signal separation (BSS) and hard decision (HD) based compensators proposed earlier by the authors. In addition to algorithm developments, big emphasis in the paper is on performance measurements and verifications carried out using measured true-world receiver front-end signals obtained from a laboratory homodyne measurement system. In general, the obtained results indicate that I/Q imbalance can be efficiently compensated using the proposed ideas. Even with really high-order modulation techniques, such as 64-QAM, close-to additive noise bound compensator performance is demonstrated using the measured signals