The Information Age is far from slowing down. Along with the explosion in wire-
lessly connected devices, a giant amount of data is being generated at a pace that we never
witnessed before. Not only this information can be stored, but also streamed for numerous
applications. All these changes in the paradigm demanded new approaches for wireless com-
munications. In the recent studies, Massive Multi-Input Multi-Output systems (mMIMO), with
hundreds or thousands of communicating antennas, are the most prevalent candidates to the
evolution of wireless communications.
Despite these systems' advantages, the joint optimization of spectral and energy e±ciencies is only
achieved by combining multilevel modulations and e±cient power ampli¯cation. To reach this
target, a transmitter employing high order modulations can spread the information into several
ampli¯cation branches, which will process only the result of the decomposition of the multilevel
constellation symbols into quasi constant envelope signals. The decomposed signals are then sent
in parallel, assuming a fully connected architecture, where each Radio Frequency (RF) chain is
connected to a single antenna. The original constellation symbol will be re-generated over the
ai, so this composition can be adjusted according to the direction in which the antenna array is
optimized. The performance of this architecture is studied, comparing the adoption of symbol
decomposition using Binary Phase Shift Keying (BPSK) components or Quadrature Phase Shift
Keying (QPSK) components.
Additionally, the high number of antennas involved in this type of communication causes the
channel matrix size to increase, augmenting the complexity of the equalization process, which can
increase power consumption and latency. We studied the combination of a multi-layer transmitter
with a low complexity receiver based on an Iterative Block Decision Feedback Equalizer (IBDFE).
This receiver avoids the matrix inversion operation in the equalizer feed-forward by replacing it
with an Equal Gain Combiner (EGC) or a Maximum Ratio Combiner (MRC) module. Results
show that this architecture can be used without penalties on performance, provided that the
number of communicating antennas is high.