Exjobbspresentation: User-Centric Distributed MIMO for LEO Non-Terrestrial Networks: Resource Optimization and Doppler-Resilient Waveform Design

Exjobbet genomfördes på Ericsson AB med Gang Zou som handledare, Aleksei Fedorov som akademisk handledare och Michael Lentmaier som examinator. 

Non-terrestrial networks (NTNs) are expected to become an important component of future sixth-generation (6G) wireless systems by extending broadband coverage to remote, maritime, aerial, and underserved areas. Among different NTN platforms, low Earth orbit (LEO) satellite constellations are particularly attractive due to their lower propagation delay compared with geostationary systems. However, the high orbital velocity of LEO satellites and the long satellite-to-ground propagation distance introduce severe physical-layer challenges, especially in high-frequency downlink transmission.

This thesis investigates downlink user-centric distributed multiple-input multiple-output (UC-D-MIMO) for LEO NTNs, where several spatially separated satellites jointly serve one ground user. Compared with conventional single-satellite transmission, UC-D-MIMO can provide macro-diversity, improved service continuity and additional spatial degrees of freedom. However, two major impairments limit these gains. First, the different line-of-sight velocities of the cooperating satellites create differential Doppler and residual carrier-frequency offset (CFO), which may destroy orthogonal frequency-division multiplexing (OFDM) subcarrier orthogonality. Second, the different satellite elevation angles cause path-loss imbalance, making the effective MIMO channel ill-conditioned and reducing the exploitable spatial rank.

To address these impairments, two complementary methods are proposed. A joint strategy and satellite selection scheme (JSSS) with adaptive power allocation (APA) is developed to mitigate path-loss imbalance by selecting transmission policies that balance spectral efficiency and channel conditioning. In parallel, an adaptive subcarrier spacing (ASCS) framework with residual-CFO tracking is proposed to maintain OFDM coherence under time-varying Doppler conditions while remaining compatible with 3GPP New Radio (NR) numerology constraints. The proposed methods are evaluated using MATLAB simulations based on realistic LEO orbital geometry, Ka-band link-budget assumptions, time-varying satellite visibility, atmospheric attenuation, and NR NTN waveform parameters. The results show that distributed satellite transmission can improve the effective MIMO rank and spectral efficiency compared with a collocated single-satellite baseline, provided that path-loss imbalance and differential Doppler are properly controlled. JSSS/APA improves the robustness of the effective MIMO channel, while ASCS identifies a Doppler-robust operating point that preserves synchroinization without unnecessary loss of goodput. Overall, the thesis shows that UC-D-MIMO is a promising architecture for LEO NTN downlink communication, but its practical benefit depends on joint geometry-aware resource allocation and Doppler-aware waveform design.