Widely linear filtering in modern MIMO transceivers

This thesis deals with the multiple-input multiple-output (MIMO) equalization with widely linear (WL) filtering. The WL filtering outperforms the linear one in presence of rotationally variant signals, with no significant complexity increase. It has been used in order to compensate the transmitter and receiver in-phase and quadrature (IQ) imbalance and to design the optimum constellation transmitted over a channel known at the receiver, according to minimum mean square error (MMSE) criterion. Moreover, with reference to OFDM-OQAM systems, it has been proposed an ad hoc MMSE WL-DF equalizer which performs a selective MIMO equalization according to the characteristics of the MIMO channel describing the system. The synthesis of the equalizer has been carried out by an algorithm which exhibits minimum storage requirements. The equalizer has proved efficient for OFDM-OQAM systems, achieving the best trade-off between performance and complexity.