Current research in the DMIMO precoding framework is mostly based on linear algorithms (e.g. ZF, MMSE), where the relation between the input data and the output transmit signal vector is a simple linear filtering based on a precoding matrix. Nonetheless, it has been widely shown in the massive MIMO literature that resorting to non-linear precoding can considerably enhance the sum rate performance, allowing to exploit the available resources more efficiently. Examples of non-linear precoding schemes are Tomlinson-Harashima (TH) and vector perturbation (VP) precoding, as well as symbol-level precoding (SLP) . While the TH and VP schemes perturb the input data symbols in order to efficiently cancel out the multi-user interference, SLP algorithms, developed more recently, optimize the transmit signal with the aim to constructively exploit the multi-user interference.
The objective of this thesis project is to perform a literature review of existing non-linear precoding algorithms in the centralized MIMO framework, and to choose a relevant scheme to implement so as to assess numerically its performance gain and complexity with respect to linear precoding. In this context, an effort towards an extension of the considered non-linear precoding scheme to a DMIMO framework is encouraged. This direction is particularly attractive in current research, since it has the potential of combining the DMIMO gains with the performance enhancement offered by non-linear processing.
- Master student in Electrical Engineering, Computer Science or equivalent.
- Experience in modeling, link level simulation and optimization.
- A solid theoretical background in areas such as information theory and signal processing.