Fix-Point Representation of a Properness-Based Algorithm for Blind I/Q Mismatch Compensation

Abstract

To meet the growing requirements for advanced wireless communications systems, a high flexibility of all subsystems is necessary. In order to keep costs low the specifications for the analog components can be reduced, if the compensation of the resulting impairments is done by digital signal processing. An example for such impairments are slight differences between the I- and Q-branch low pass filters due to production tolerances in direct-conversion receivers that lead to frequency selective I/Q imbalance. A waveform independent digital compensation approach based on second order statistics shows the desired flexibility. Without the knowledge of any transmit data the I/Q mismatch can be compensated by adaptive filtering. This is done by the restoration of a statistical property, called properness, which has been disturbed by the I/Q mismatch. It is shown, that for LTE standard compliant transmission signals under real world conditions the algorithm is ideally suited in terms of performance and hardware implementation. In a fix-point implementation the approach provides a normalized mean square error of below-60 dB, compared to the floating-point simulation.