Comparison of Purely MPR-Based and IRSA Random Access: Spectral Efficiency Versus Energy Expenditure

Grant-free random access enables massive Internet of Things (IoT) connectivity in Non-Terrestrial Networks (NTNs), where a large number of low complexity devices intermittently transmit short packets under tight energy and reliability constraints. Repetition based coded random access schemes, such as Irregular Repetition Slotted ALOHA (IRSA), achieve near-optimal throughput under idealized collision channel assumptions, while successive interference cancellation based multi-packet reception schemes attain high spectral efficiency under more realistic physical-layer models. Nevertheless, a consistent comparison between these approaches under unified reliability and energy constraints is still missing. This paper provides a unified analytical and simulation based comparison of MPR Adaptive Slotted Access (MASA) and IRSA under a common physical-layer model, using the same frame structure, channel assumptions, and packet-level reliability constraint. The analysis explicitly links packet-level reliability to replica-level decoding in repetition-based access and captures the resulting impact on required decoding thresholds and energy expenditure. Results show that although IRSA can asymptotically approach the spectral efficiency of adaptive multi-user decoding, doing so under a common outage constraint entails a rapidly increasing energy cost. In contrast, MASA maintains high spectral efficiency with a significantly lower energy expenditure by exploiting multi-user decoding over the full data interval, thereby delineating when the throughput gains of coded random access remain energy-efficient in non-terrestrial massive IoT systems.