This is part one of a two part blog about our SD-RAN and SD-Core open source projects. This blog provides an overview of the current status of SD-RAN and SD-Core through 2021; the second blog offers insights into plans for each of these projects going forward.
Background
ONF has been developing disaggregated, software-defined, virtualized and cloud-based open-source platforms for RAN and mobile core under its SD-RAN and SD-Core projects, respectively, for a few years now.
ONF’s SD-RAN project is building open-source components that complement O-RAN’s focus on architecture and interfaces for disaggregation, SDN and cloud. Similarly, SD-Core is ONF’s flexible, agile, scalable, and configurable dual-mode 4G/5G core network platform. Both of these platforms are optimized for a hybrid cloud deployment and provide SDKs/APIs enabling programmatic telemetry collection and control and configuration by 3rd party applications and platforms.
With the advent of 5G, mobile networks increasingly need to support denser deployments, and simultaneously serve applications that are much more diverse and demanding. Furthermore, this diversity and demand seems to be ever evolving, requiring the network to be able to continuously evolve in parallel. As the applications become more and more mission critical, the mobile network also needs to be much more resilient against potential failures and cyber attacks. And of course, all of this needs to happen at an attractive TCO for the mobile network operator.
The mobile network architecture that can address these requirements needs to be built on the pillars of disaggregation, virtualization, SDN, and cloud, across the RAN and the mobile core. The network needs to be top-down and end-to-end programmable, open, automated, built on white box hardware and merchant silicon, cloud centric and cloud operated, secure and multi-tenant.
SD-RAN Accomplishments
ONF’s SD-RAN project is building open-source components that complement O-RAN’s focus on architecture and interfaces for disaggregation, SDN and cloud. The goal is to foster the creation of true multi-vendor RAN solutions with open interfaces and open source components, helping invigorate innovation across the RAN ecosystem.
At the center of the project is the development of an open source near real-time RIC (nRT-RIC). SD-RAN’s nRT-RIC is based on a cloud-native and microservices-based and RAN-optimized version of ONF’s production-grade SDN controller, ONOS. SD-RAN’s nRT-RIC provides O-RAN interface terminations, R-NIB, UE-NIB, clustering, configuration, subscription and service model support services in its microservices architecture.
The SD-RAN project also facilitates the development of several open source and proprietary xApps running on this RIC. These xApps leverage O-RAN specified E2 Service Models and also new pre-standard Service Models, which have been developed as part of the SD-RAN project. Embracing the software-centric, and cloud-based approach, an SDK-approach has been endorsed by the project as the interface between the nRT-RIC and xApps.
The RAN E2 nodes (RU, DU, CU) to be controlled come from the project ecosystem, but ONF also provides an LTE white-box open-source reference CU/DU software and hardware integrated with our nRT-RIC. Furthermore, the SD-RAN project has also developed a RAN Simulator, to quickly develop and test new xApps, and new Service Models, with the aim of fueling innovation. Finally, the SD-RAN end-to-end solution can also be experienced (for dev/test) entirely in a virtualized form, in a VM or server, using RiaB (sdRAN-in-a-Box). The entire stack including xApps, RIC components, CU/DU, UE and mobile-core can be instantiated within RiaB with a few simple commands.
As part of the SD-RAN project, ONF hosts a Joint Task Group where ONF partner operators (AT&T, China Mobile, Deutsche Telekom) and ONF get together, strategize how O-RAN specifications should evolve, discuss new ideas, and form consensus to jointly contribute contributions to O-RAN Alliance.
SD-RAN platform is being released with an ONF member-only license with regular cadence. The latest member-only release was v1.3 in Oct 2021. As announced earlier this year, trailing releases are being open-sourced with Apache 2.0 license. Consequently, SD-RAN v1.2 is now open-source.
Since its inception, the SD-RAN project has gained significant attention and support from the ecosystem. In 2021 the SD-RAN community demonstrated several “firsts” and key innovations in the on-going SD-RAN outdoor trial at Deutsche Telekom, that showcases the feasibility of the multi-vendor, multi-cloud-based O-RAN architecture operationalization. Specific features include:
- Fully disaggregated O-RAN compliant 4G and 5G RAN,
- Multi-vendor disaggregated 5G small cells built to run on commodity hardware,
- Open source RAN Intelligent Controller (RIC) controlling both 4G and 5G networks simultaneously using SDN principles,
- Concurrent operation of multiple xApps from different vendors,
- Cloud hosted, cloud managed, containerized, 3GPP Release 15 compliant 4G and 5G core network, with support for network slicing using multiple User Plane Functions (UPFs)
- Top-down and end-to-end programmable 4G and 5G network providing rich end-to-end network visibility.
Additional SD-RAN accomplishments in 2021:
- SDK-Development: Architectural development and implementation of Go and Python based SDKs from ONF and Facebook.
- SON Service Model and xApp Development: Development of an xApp that links AirHop’s eSON platform to nRT-RIC as well as the development of a pre-standard RAN Control Service Model that enables it, with joint collaboration between ONF, Facebook, AirHop and Radisys
- Integration with Commercial Disaggregated Small Cells and 3rd Party xApps: End-to-end operationalization with the Radisys 5G SA L2/L3 stack, and AirHop/Facebook SON xApps.
- Open-Source xApps: Development of xApps for RAN telemetry extraction, and PCI conflict mitigation, Mobility Load Balancing & Handovers, and finally RAN Slice management.
The ONF SD-RAN team is starting engagements or is already actively engaged with Intel-Labs, RIMEDO-Labs, Lockheed Martin, Sercomm, Capgemini and Amdocs in addition to the participants in the SD-RAN DT Trial. The community meets twice every month at the SD-RAN Community Call and actively participates in the sdran-dev slack-channel and mailing list.
SD-Core Accomplishments
SD-Core is ONF’s flexible, agile, scalable, and configurable dual-mode 4G/5G core network platform. SD-Core builds and enhances on ONF’s 4G OMEC as well as free5GC core network platforms to create a dual-mode core that supports LTE, 5G NSA and 5G SA services. SD-Core includes two user plane function (UPF) solutions: a P4-based UPF that is realized on programmable forwarding elements allowing for fine-grained visibility for verifiably performant and secure operations, and a scalable, containerized UPF.
SD-Core is optimized for a hybrid cloud deployment. Its centralized dual-mode mobile core control plane has been designed to control many UPFs distributed across many edge clouds around the world. SD-Core provides a rich set of APIs for runtime configurability of each of its services, including its subscriber management via potentially third-party applications. These APIs also provide extensive telemetry capabilities that enable monitoring, logging, and alerts as well as verification and closed-loop control solutions around them. SD-Core can be configured at runtime for dynamically programmable network slicing, subscriber management, QoS and policy management, and access control.
SD-Core platform is released under a member-only license with regular cadence. The first release, SD-Core v1.0 was released in Dec 2021 and includes the following features:
- 5G SA Mobile Core Development: Development and hardening of the fully disaggregated, service-based and containerized 5G mobile core.
- 5G NSA support in SD-Core: Operators can use SD-Core deploy 5G NSA. 5G NSA uses 4G control plane with 5G RAN
- P4 based 4G/5G dual-mode UPF Development: Development of a P4 programmable 4G/5G dual-mode UPF architecture that enables packet level processing on the edge cloud fabric with corresponding control via an ONOS application and a PFCP agent.
- Containerized 4G/5G dual-mode UPF Development: DPDK-accelerated containerized UPF development. This architecture also enables an interface service that interconnects the UPF pod with the 4G and 5G mobile core control planes via a PFCP agent.
- 5G gNB simulator which simulates 5G UE & 5G gNodeB. This simulator is capable of supporting multiple UEs and multiple gNodeBs. Tool also capable of doing data tests. This assists quick development of the features. This tool can be used to run automated tests for the 5G core network.
- Extensive policy configuration: Provides ability to control runtime behavior of the subscriber through policy framework. Policy framework assigns various QoS parameters to subscriber sessions.
- Application Filtering: Provides ability to grant access to applications based on Slice Configuration.
- APIs Development for Runtime Configurability and Control: Programmatic configurability of subscriber management, network slicing, QoS management and access control via APIs.
- SIM Management or UE subscription management application interface: Allows third party applications to interface with SD-Core.
- SD-Core operationalization in the Deutsche Telekom Field Trial: Interoperability enablement of SD-Core’s 5G mobile core with the commercial Radisys 5G RAN stack, and ensuring broadband connectivity with commercial 5G mobile phones.
================================
To learn more about what’s coming next in 2022 and beyond, check out the next blog.
