PROJECT TITLE :

Low-Complexity MU-MIMO Nonlinear Precoding Using Degree-2 Sparse Vector Perturbation

ABSTRACT:

Multiuser multiple-input multiple-output (MUMIMO) nonlinear precoding techniques face the matter of poor computational scalability to the scale of the network. During this paper, the elemental problem of MU-MIMO scalability is tackled through a unique signal-processing approach, that is termed degree-2 vector perturbation (D2VP). Unlike the traditional VP approaches that aim at minimizing the transmit-to-receive energy ratio through searching over an N-dimensional Euclidean house, D2VP shares the identical target through an iterative-optimization procedure. Every iteration performs vector perturbation over two optimally selected subspaces. By this implies, the computational complexity is managed to be within the cubic order of the dimensions of MU-MIMO, and this mainly comes from the inverse of the channel matrix. In terms of the performance, it's shown that D2VP offers comparable bit-error-rate to the sphere encoding approach for the case of tiny MU-MIMO. For the case of medium and giant MU-MIMO when the sphere encoding will not apply because of unimplementable complexity, D2VP outperforms the lattice-reduction VP by around 5-ten dB in Eb/No and ten-50 dB in normalized computational complexity.