IQ Imbalance Mitigation via Unbiased Training Sequences

Modern digital communication systems use quadrature modulation (IQ) such as QAM. An imbalance between the I and Q paths in the RF module results in self-interference that affects both channel estimation and data demodulation. We propose unbiased training sequences that automatically eliminate the interference from the channel estimate. Strong gains are achieved for high data rates. For additional gains, we also propose a relatively inexpensive extension to compensate for data distortion in OFDM. End-to-end (transmitter and receiver) and frequency-dependent imbalances are covered.