Towards Energy Efficient 5G vRAN Servers
summary
RENC adjusts CPU P‑states to use sub‑second workload variation, coupling MAC layer rate limiting with ordered frequency changes so throttling happens before lowering clocks. Slack in real‑time vRAN threads is measured via eBPF slack hooks or minor binary rewriting, then an iterative controller converges to a target 10% slack. Tests on commodity telco servers and a two‑cell 11‑UE testbed show up to 45% CPU power reduction (about 29–30% server‑level) with no visible throughput or latency loss.
AI-written summary · about
chapters (13)
- 0:00Energy efficient 5G vRAN servers
- 2:23Network underutilization at peaks
- 3:49vRAN power consumption challenges
- 5:57Energy saving knobs in vRAN
- 7:18Rank system for power optimization
- 9:01Measuring slack in RAN software
- 10:26CPU frequency adjustment
- 11:31Handling traffic spikes in vRAN
- 12:35Open interfaces for deployment
- 13:39Experimental setup and evaluation
- 14:4530% power savings with Rank
- 15:52Performance impact and benefits
- 16:13Key takeaways and contributions
related talks
26mPower Management for Telco Operators
16mQCON: Seamless QoE-Aware 5G Streaming via Multi-Connectivity
17mEdgeRIC: Empowering Real-time Intelligent Optimization and Control in NextG Cellular...
18mEnabling SLO-Aware 5G Multi-Access Edge Computing with SMEC
14mRDC: Energy-Efficient Data Center Network Congestion Relief with Topological
17mDesigning Next Generation 5G Wireless Infrastructure Using RISC-V
17mChannel-Aware 5G RAN Slicing with Customizable Schedulers
18mA Paradigm Shift in 5G Deployment: The Promise of eBPF for Open...