Premium
On quasi‐optimum iterative DF detection for large‐scale MU‐MIMO systems
Author(s) -
Torres Paulo,
Gusmao Antonio
Publication year - 2018
Publication title -
transactions on emerging telecommunications technologies
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.366
H-Index - 47
ISSN - 2161-3915
DOI - 10.1002/ett.3260
Subject(s) - computer science , mimo , algorithm , single antenna interference cancellation , equalization (audio) , detector , iterative method , telecommunications link , channel (broadcasting) , control theory (sociology) , telecommunications , decoding methods , control (management) , artificial intelligence
This paper deals with single‐carrier/frequency‐domain equalization for uplink transmission within a broadband multiuser multi‐input–multi‐output system, where a large number of base station antennas ( N R ) can be adopted, possibly much larger than the number of transmitter antennas ( N T ) jointly using the same time‐frequency resource at mobile terminals. In this context, we propose low‐complexity iterative decision‐feedback (DF) detection techniques, which can be an interesting alternative to the usually recommended linear detection techniques. Our performance results for a range of quadrature amplitude modulation schemes are discussed with the help of selected performance bounds, with a special attention being devoted to the convergence of the iterative process and to the impact of an imperfect channel estimation. They confirm that simple linear detection techniques, designed to avoid the need of complex matrix inversions, can lead to unacceptably high error floor levels unless N R ≫ N T . However, it is shown that the combined use of such simple linear detectors and suitable interference cancellation procedures within the proposed class of iterative DF detection techniques can offer a nonnegligible performance advantage over the somewhat more complex (due to the required matrix inversions) linear minimum mean‐squared error detection technique. Moreover, we can achieve close approximations to the appropriate performance bounds, which are able to take into account channel estimation imperfections, even forN RN T< 10 , provided thatN RN Tis also high enough to ensure iterative DF convergence.