User Grouping in MU-MIMO Air to Ground Communications

Kurzfassung

Air-to-ground communications must support increasing communication demands while operating over limited spectrum resources. Uplink multi-user multiple-input multiple-output (MU-MIMO) can increase resource reuse by allowing several single-antenna aircraft to transmit simultaneously over the same time-frequency resource and exploiting the multiple antennas at the ground station to separate their signals. Performance, however, depends strongly on which aircraft share each resource, since the joint structure of their channels determines the multiuser interference observed by the receiver. User grouping is therefore a relevant resource allocation problem, while exhaustive optimization is impractical because the number of possible partitions grows combinatorially with the active user population. This Master’s Thesis investigates low-complexity user grouping for uplink air-to-ground (A2G) MU-MIMO communications, using outage probability at a prescribed target spectral efficiency as the main performance metric. An empirical descriptor analysis determines which channel, geometrical, and cluster balance properties are informative about the outage ordering of candidate partitions. The analysis identifies LogDet as the most consistent descriptor of grouping quality and motivates two constructive algorithms, Pair-wise LogDet and Clusterwise LogDet, which represent channel interactions at different levels. The proposed methods are evaluated in controlled single ground station scenarios with different active user populations, target users per cluster, and target spectral efficiencies, and compared with random grouping and adapted literature methods. Their robustness is further examined with outdated aircraft positional information and under successive interference cancellation (SIC) reception. The analysis is also extended to co-located and geographically distributed receiver architectures. Finally, a temporal framework based on grouping reuse, repair, and bounded refinement is evaluated over a 24-hour historical OpenSky traffic sequence with aircraft arrivals and departures. The results show that channel informed grouping substantially reduces outage relative to random allocation and that Clusterwise LogDet consistently matches or improves upon Pairwise LogDet. The proposed grouping principles remain effective under moderate positional delays and SIC reception, while geographically distributed reception provides additional spatial diversity compared with a co-located receiver using the same total number of antennas. Over successive historical traffic snapshots, previously accepted groupings can frequently be reused, and repair handles changes in the active aircraft population at much lower computational cost than constructing a new grouping from scratch. Bounded temporal refinement provides additional outage reductions, particularly when instantaneous grouping performance degrades. This Master’s Thesis was carried out during an Erasmus mobility period at the Institute of Communications and Navigation of the German Aerospace Center (DLR) in Oberpfaffenhofen, Germany.