Intelligent auto-scaling in 5G RAN based on MicroK8s with Ubuntu real-time kernel on Intel® technology
Communications service providers (CoSPs) must perform a triple balancing act: handle ever-increasing amounts of network traffic generated by millions of devices while still meeting high availability service-level agreements (SLAs) and reducing operational costs (OPEX). As the 5G network matures, timeliness of response and scalability are increasingly important, whether it’s supporting a stadium full of fans and complex communications equipment, smart supply chains, autonomous vehicles, advanced healthcare use cases or Industry 4.0.
Intel and Canonical are committed to helping the telecommunications industry realise the potential of 5G—increasing revenue for CoSPs and improving the user experience for consumers. Together, the two companies have collaborated to propose an architecture and framework that supports intelligent auto-scaling of the DU in a FlexRANTM-based 5G testbed, using general-purpose hardware. The deterministic response times delivered by Canonical’s real-time kernel make Ubuntu 22.04 LTS the ideal OS for Intel FlexRAN solutions. Deployments that need a small footprint can take advantage of MicroK8s, while general-purpose Intel® hardware and AI algorithms built on Intel® oneAPI provide cost efficiency and scalability.
Ideal to ensure SLA compliance and reduce expenditures
Intel and Canonical co-validated the FlexRAN software development kit (SDK) on Ubuntu 22.04 LTS with RT kernel, enabling FlexRAN-based 5G E2E deployments that require preemptive RT kernel capabilities to meet 5G latency requirements. Running FlexRAN on Ubuntu allows CoSPs to use familiar upstream Kubernetes, automation frameworks and open-source tools, and see performance gains from the latest real-time kernel package tuned for telco.
Architecture features include:
- Lightweight containerised environment
- Better underlying Ubuntu performance, with average latencies for GNB_DL_LINK and GNB_UL_LINK decreased by 17.5% and 4.4%, respectively.
- Auto-scaling efficiency
- Deterministic response times
- Intelligent traffic prediction
Download the whitepaper to learn more about the solution and the performance results.