The Network MIMO Downlink with Per-Antenna Power Constraint: Linear Preequalization vs. Tomlinson-Harashima Precoding

Preequalization/precoding schemes for use in network MIMO downlink scenarios are assessed and compared. Whereas conventional linear preequalization and Tomlinson- Harashima precoding (THP) only obey a sum power constraint, in the situation at hand, a per-antenna power constraint is more meaningful. We evaluate the power statistics of classical transmission schemes, which are only designed according to a sum power constraint. Next, we review the optimization of linear preprocessing with per-antenna power constraint and extent the procedure to the optimization of THP. The incorporation of such a restriction into lattice-reduction-aided (LRA) precoding is also addressed. In all cases, the performance of the schemes is covered by means of numerical simulations for a typical network MIMO setting coordinating three base stations and three users. It is shown, that non-linear precoding schemes achieves gains in performance at even lower per-antenna requirements.