Generalized Pre-coding Aided Spatial Modulation (GPSM) is a MIMO system communication strategy in which the transmitter encodes the information in two entities: the parallel transmission of symbols belonging to a digital modulation constellation, and the choice of the indices of the information vector elements that carries the informaiton symbols, denoted information bearing positions (IBP), while the remaing positions are set to zero. Besides, the transmitter precodes, prior to transmission, the information vector, which lets the receiver node benefit from complexity reduction. Among the advantages of this strategies, the concentration of the transmitted energy only on the information vector positions that transports modulated symbols, resulting in system performance improvement. This thesis considers a multiuser MIMO (MU-MIMO) that employs GPSM in the donwlink transmission. The presented signal model to describe this system evidences the influence of the IBP coding matrices on the system performance. Based on this model, GPSM MU-MIMO system is presented considering three linear precoders: Zero-Forcing, Block Diagonalization and Double-Stage Block Diagonalization. For each precoder, strategies to select the IBP encoding matrix, acording to the matrix that describes the current MU-MIMO channel, is proposed. These matrices are to be employed by the user, aiming at system detection performance improvement. Detection performance curves are compared to semianalytic lower bounds. Finally, a scenario in which that are a number of available antennas at the BS and/or at the users that exceed the number of radiofrequency chains. this thesis porposes optimal and reduced complexity strategies that exploit the use of the most favorable antennas for transmission and/or reception, in addition to the choice of the IBP enconding matrices, aiming at system performance improvement.