Discreteness-aware Receivers for Overloaded MIMO Systems
share
We describe three new high-performance receivers suitable for symbol detection of large-scaled and overloaded multidimensional wireless communication systems, which are designed upon the usual perfect channel state information (CSI) assumption at the receiver. Using this common assumption, the maximum likelihood (ML) detection problem is first formulated in terms of an l0-norm-based optimization problem, subsequently transformed using a recently-proposed fractional programming (FP) technique referred to as quadratic transform (QT), in which the l0-norm is not relaxed into an l1-norm, in three distinct ways so as to offer a different performance-complexity trade-off. The first algorithm, dubbed the discreteness-aware penalized zero-forcing (DAPZF) receiver, aims at outperforming state-of-the-arts (SotAs) while minimizing the computational complexity. The second solution, referred to as the discreteness-aware probabilistic soft-quantization detector (DAPSD), is designed to improve the recovery performance via a soft-quantization method, and is found via numerical simulations to achieve the best performance of the three. Finally, the third scheme, named the discreteness-aware generalized eigenvalue detector (DAGED), not only offers a trade-off between performance and complexity compared to the others, but also differs from them by not requiring a penalization parameter to be optimized offline. Simulation results demonstrate that all three methods outperform the state-of-the-art receivers, with the DAPZF exhibiting significantly lower complexity.
READ FULL TEXT