Distortion-Aware Power Allocation for 6G MIMO Transmitters

Sammanfattning

Future sixth-generation (6G) wireless systems are expected to support extreme data rates, massive connectivity, and high energy efficiency. Largescale Multiple-Input Multiple-Output (MIMO) combined with Orthogonal Frequency Division Multiplexing (OFDM) is a key enabling technology to meet these demands. However, OFDM signals exhibit a high Peakto- Average Power Ratio (PAPR), which forces power amplifiers (PAs) to operate with significant back-off in order to satisfy Error Vector Magnitude (EVM) and spectral emission constraints, thereby reducing transmitter efficiency. This thesis investigates distortion-aware power allocation in OFDMbased MIMO transmitters with a focus on 6G-oriented system design. The work analyzes the spatial and spectral behavior of distortion introduced by Crest Factor Reduction (CFR) and nonlinear PA characteristics. A comprehensive simulation framework is developed based on wideband CP-OFDM transmission, DFT beamforming, and multi-user MIMO to evaluate distortion under spatial, frequency, and joint spatial–frequency multiplexing. The results reveal that branch-level distortion does not directly translate to user-experienced distortion. With increasing number of simultaneously served users, nonlinear distortion becomes progressively decorrelated from the desired signal components, resulting in significantly lower user EVM compared to branch EVM. A similar decorrelation effect is observed when users are separated in the frequency domain through sub-band allocation. Furthermore, increasing the number of antennas enhances distortion averaging, leading to improved user-side signal quality while branch-level distortion remains largely unchanged. Building on these observations, the study introduces the concept of power overbooking, whereby additional transmit power can be allocated without violating user-level EVM constraints. The results demonstrate that joint exploitation of spatial and frequency resources provides measurable power headroom gains, enabling reduced PA back-off and improved energy efficiency. Overall, this work highlights that distortion-aware and hardware-aware resource allocation can significantly improve transmitter performance. The findings provide practical design insights for future large-scale MIMO systems, where exploiting distortion decorrelation across space and frequency becomes a key enabler for energy-efficient 6G transmitter architectures.