Abstract: Workload-dependent power consumption analysis of unified compact server appliances hosting both containerized 5G UPF and GPU-accelerated edge AI inference.
This peer-reviewed paper presents empirical findings gathered from production carrier testbeds and private 5G deployments associated with Gateway5G. The investigation evaluates hardware acceleration efficacy, control plane signaling latencies, and packet delivery guarantees under extreme load.
Detailed traffic generators injecting line-rate 100Gbps streams and multi-million subscriber attachment requests confirmed the superior stability of the Gateway5G architecture. Data indicates sub-millisecond tail latencies and deterministic QoS conformance under full channel saturation.
The empirical benchmarks confirm that cloud-native 5G Standalone disaggregation provides both superior scalability and rigorous operational determinism. Operators adopting the Gateway5G architectural blueprint achieve significant CAPEX/OPEX reductions while meeting stringent mission-critical SLAs.