5G has the potential to deliver a new generation of services thanks to higher data rates and ultra low latency. To take advantage of this, communications service providers (CSPs) are looking to move data processing and compute power closer to the end user. The end user could be anything from a smartphone, to a driverless car, or a robot on a production line - all will form part of a fully software-defined network. This puts both edge computing and cloud-native capabilities at the centre of CSP plans to turn 5G networks into revenue-generating platforms for new services, including vertical market opportunities. It also paves the way for wider societal benefit through more efficient technological and human resource utilisation.
Services Based Architecture
5G standardisation has taken a novel approach and focused on services, software, standard APIs and security model. Part of the journey to a services based architecture is the move from virtual machines (VMs) to cloud-native network functions. This relies on the modernisation of existing applications, maturation of Kubernetes operational tooling, and continued evolution of OSS/BSS.
At the edge, network functions virtualization (NFV) is already being deployed at aggregation sites. Here, CSPs are using OpenStack with distributed nodes for software defined-wide area networks (SD-WAN) and mobile applications. The next step will be to use cloud-native applications on Kubernetes-based container platforms like Red Hat OpenShift for 5G Core (5GC) and radio access networks (RAN). This rides on the availability of 5G spectrum, maturation of applications, and volume economics.
Network slicing technology is in the demo phase, with radio and 5GC vendors exploring how it can be achieved from RAN to core. Open interfaces and protocols to bring these capabilities to life across shared 5G RAN – multi-vendor and multi-operator – are under development. One avenue being looked at for service-based slicing end-to-end is using artificial intelligence (AI) to help automate operational decision-making for the overall cloud-native infrastructure.
Containers
A containerised core will be the heart of cloud-native 5G networks. Managing and scaling networking apps in containers using a modular microservices approach will help service providers to dynamically orchestrate and grow service capacity across a distributed architecture.
Open APIs will enable each network function to connect and interact with other network functions, helping expand a CSP’s use of virtualisation beyond centralised core operating sites and out to the enterprise edge, including the radio access network (RAN) application sites they operate. If the entire technology stack – application, network and data platforms – is built on open source applications and services, connected through open APIs, service providers will gain flexibility, consistency and security right across their architecture.
To make this open network a reality, CSPs will want to consider a common infrastructure running across the organisation. Some have already implemented such a horizontal cloud platform, and are seeing accelerated time to market and reduced operational costs. Turkcell has done so, and has cut launch time for new services by 66%, and is gaining return on investment three times faster.
Cloud economics
Moving to the cloud is often associated with efficiency gains and cost savings, but there are differing strategic approaches to take, and operators are weighing up which will drive most revenues with the least risk. To put some numbers behind the options, Red Hat used ACG Research’s Business Analytics Engine to model tier 1 operator networks. It found that an open, integrated horizontal cloud resulted in 40% lower operational expenditure (opex) and 30% lower total cost of ownership (TCO) over five years as compared to vertically integrated (silo-based) solutions. (See: Economic advantages of virtualizing the RAN in mobile operators' infrastructures).
The study also examined radio access networks (RAN), as they are critical to customer experience, yet typically expensive, requiring specialised software and hardware that can be difficult to upgrade and scale. To assess the business value of virtualizing the RAN, Red Hat used ACG Research’s ‘Business Analytics Engine’ to compare vRAN TCO against a conventional distributed RAN (DRAN), in which radio units (RU) and baseband units (BBU) are placed at each radio site. Comparing each design over five years, vRAN capex was found to be half that of DRAN. This was due to the lower cost of servers and related infrastructure that is needed for BBU processing at a smaller number of server sites, compared with the cost of BBU equipment installed at each cell site in the DRAN design. Opex was found to be 40% lower for vRAN than DRAN, due to smaller footprint, lower site rental and fiber lease, less BBU maintenance including lower power and cooling costs. Overall, up to 44% in TCO savings are projected if the entire 4G/LTE RAN is virtualized, along with the growth of new and existing cell cites.
The agile advantage
Service providers are looking at whether to carry vRAN deployments in existing 4G/LTE networks forward into 5G networks, or to adopt a greenfield approach, as Japan’s Rakuten Mobile has done. The company has been in the spotlight for creating a fully cloud-native mobile network based on open software and open standards to boost its agility and innovation, helping differentiate it from legacy mobile vendors as the world moves to 5G.
Open source is becoming the foundation for the end-to-end cloud-native networks of tomorrow. It provides freedom and flexibility to choose technology from different vendors, helping reduce risk, and it galvanises industry-wide collaboration, speeding up innovation and standardisation. An open deployment enables CSPs to expand containerised and cloud-native architectures beyond their networking environments, across their operations and data applications, at the edge, and beyond into enterprise networks. This means CSPs will be better placed to embrace innovative software to support new services such as IoT or augmented and virtual reality for diverse use cases.
Given that the industry must have trusted software sources for reliability, security and support, we are seeing players from the vendor and system integrator ecosystem partnering more openly and deeply than ever before, to better address changing real-world needs. The road to 5G may see the most collaborative and participatory effort so far in the history of telecommunications.
