5G necessitates greater levels of adaptability in telecom network architecture, scaling, and implementation. This is especially true for Radio Access Network (RAN) deployments, where cloud technology provides new, innovative options that supplant tried-and-proven, purpose-built solutions. The term “open RAN” refers to an open radio access network architecture with open, interoperable interfaces that promotes big data and AI-driven improvements in the RAN area. Cloud Native RAN is the consequence of a large-scale implementation of cloud technologies with web-scale principles.
5G RAN includes all hardware and software combinations (architectures [DRAN, CRAN, oRAN, vRAN], spectrum bands, and form factors) of NR, as well as all software pertaining to carrier aggregation, dynamic spectrum sharing, and remote baseband unit (BBU) upgrades from LTE to 5G. 5G core functions include user plane functions (UPF), control plane functions, policy administration, and subscriber data management etc.
Detailed coverage of 5G Architecture is here –
https://www.vamsitalkstech.com/5g/5g-core-5gc-platform-architecture/
https://www.vamsitalkstech.com/5g/real-world-5g-deployment-architecture/
We’ve witnessed the telecom industry’s networks move from a physical bare-metal system to virtual network functions (VNFs). Most businesses now run container network services under network function virtualization. However, industry norms are constantly changing. As they transition to 5G networking, telecom is changing to function decomposition (such as Open-RAN and Control-User Plane Separation), distributed edge (due to increased capacity need), and service-based design that employs REST APIs rather than traditional protocols. Cloud-native microservice architecture is being embraced by 5G architecture trends. Containers based on Kubernetes and NoSQL databases for stateless architecture are widely used in telecom networks. Much of 5G in the telecom sector follows this structure, including operations support systems/business support systems, core networks, and radio access networks. Another 5G design pattern is more logical and virtual network isolation. In end, all this entails supporting multiple logical network slices for all of the enterprise’s unique use cases.
Cloud RAN – Key Considerations
Cloud RAN is the implementation of RAN functionalities over a generic computational platform as opposed to a specialized (custom built hardware platform) and the management of RAN application (and virtualization) using cloud-native principles. Running a few 5G RAN network services in containers through commercial off-the-shelf (COTS) hardware platforms is the first step in cloudifying the RAN. The Centralized Unit is located closer to the control plane and user plane, while the Distributed Unit continues with latency-sensitive radio processing.
To lay the groundwork for a successful Cloud RAN implementation, we divide Cloud RAN into four parts as shown above (in red).
- Choosing the appropriate hardware platform and environment for COTS devices, including accelerators
- Utilizing cloud native technologies like Kubernetes and DevOps principles, cloud native architecture enables the realization of RAN functions as microservices in containers running on bare metal servers
- End-to-end life cycle management of services across Cloud RAN, transport, 5G core, and underlying cloud infrastructure is made possible via management, orchestration, and automation
- The deployment of non-RAN and enterprise functions in a virtualizion layer to suport business continuity
From these requirements as such, and from the need to enable the joint support of diverse use cases in an economically feasible way, the following RAN architecture requirements have been captured on high‐level in and formulated in more detail.
#1 RAN Hardware and Environment
The 5G RAN should support scale up or down based on throughput, the number of mobile devices, the number of connections and data volumes etc. To enable this, it should be able to handle and scale CP and UP individually. Note that 3GPP explicitly also envisions that CP and UP signaling as being handled by different sites, which means distributed server deployments.
The User Plane Function (UPF) moves closer to the subscriber edge is another aspect of designing an efficient xHaul network. Because UPF provides actual connectivity between the subscriber traffic and external data networks, the placement of UPF is heavily dependent on the peering availability at the installation location. When deploying 5G RAN, most CSPs will use a mix of managed and OEM hardware.
#2 Kubernetes and Cloud Native
The 5G network spec from 3GPP brings telco into the cloud-native domain i.e the network becomes a cloud that is scalable, programmable, and dynamic. Unlike 4G LTE, 5G RAN architectures split the gNB base station into Central Units (CUs) and Distributed Units (DUs). This architecture (which can be distributed based on the exact use case as we will see in the reference architectures) brings about performance management, optimization, and overall system load management. The payoff is that virtual network slices can be provisioned on-demand with business objective-driven QoS guarantees.
https://www.vamsitalkstech.com/5g/the-final-post-of-2021-telco-optimized-kubernetes/
#3 End to end lifecycle management (LCM)
RAN deployments should operate efficiently in complex 5G deployments across a wide range of deployment scenarios. Depending on where compute and network is provisioned, they should be able to maximally leverage cloud region-based processing Lifecycle Management Tools key capabilities such as instantiation and provisioning, configuration management, monitoring, governance control, and resource optimization of 5G RAN and 5G Core as well. In a hyperscaler like AWS, all orchestration and management tools are leveraged using a mix-and-match approach based on service demand and other business requirements. Service orchestration interprets service needs using intent-based interfaces, creates policies for the corresponding SLAs for the services, and sets methods for connections established across the network to support those requirements. Between the top layer and the base network is the service orchestration system. For the purpose of exposing capabilities in business terms, it abstracts the network complexity. It offers the interface for the management of service orders and orchestrates the creation of services across network domains This gives network orchestration systems an interface and executes the required operations within the relevant network domains to support the end-to-end services. Clients request digital services and network slices using a service orchestration-based approach.
#4 Accommodating legacy VNFs, MECs, and Operational Systems
While they deploy 5G networks, CSPs must still maintain a complex technology landscape that includes systems for 4G LTE, IMS, billing, provisioning, and other operations that maintain their business processes. Customers still need a proven methodology for migrating workloads and running their business to the cloud. Many CSPs must maintain legacy systems to support and ensure customer satisfaction and continuity while preparing for the rollout of 5G as well as optimizing business processes to fund innovation.
Conclusion
Designers of 5G services must offer their CSPs the flexibility to deploy and use 5G infrastructure internationally in order to achieve high availability, elasticity, and scalability requirements while also meeting strict security standards. Cost reduction, increased agility, and an optimized global presence are the main objectives.
Featured image by Z z: https://www.pexels.com/photo/5g-metal-sign-under-wire-construction-6200343/