Joint Energy–Spectral Efficiency Trade-Offs in a Multi-Slice 5G-NR Uplink: A Link-Level Evaluation with Per-Slice PUSCH Configuration

Network slicing allows a single 5G New Radio (NR) carrier to serve services with significantly different reliability targets. However, the cost of each slice in spectral and energy terms, once it is realized as a concrete uplink, is rarely measured. This paper reports a link-level evaluation of the joint energy efficiency (EE) and spectral efficiency (SE) trade-off in a multi-slice 5G-NR uplink in which enhanced Mobile Broadband (eMBB), Ultra-Reliable Low-Latency Communication (URLLC), and massive Machine-Type Communication (mMTC) each receive a distinct Physical Uplink Shared Channel (PUSCH) configuration matched to a block error rate (BLER) target of 10−3, 10−5, and 10−1, respectively. Using a 3GPP-compliant simulator built on the MATLAB 5G Toolbox (R2026a), 1440 operating points across different frequency bands, FR1 and FR2, propagation delay profiles, transmit power levels, and user distances from the gNB were evaluated. Each point is evaluated twice, with and without a co-channel fixed-service (FS) interferer. Every slice exhibits an interior energy-optimal transmit power; the three slices occupy clearly separated regions of the EE–SE plane, ordered eMBB above mMTC above URLLC. Finally, FS interference reshapes the trade-off through one mechanism with very different consequences per slice: eMBB, which earns its rate from 256-QAM, loses approximately a quarter of its peak throughput and roughly half of its peak EE under line-of-sight (LoS), while mMTC is the most resilient and URLLC shows the sharpest qualitative change, with its energy optimum migrating by approximately 16 dB. Therefore, the energy-optimal operating point is not a fixed property of the band, distance, and slice, but a function of the interference that the link actually sees; thus, slice-aware uplink power control must also be interference-aware.