Open RAN needs to automate – and fast

The case for automation is clear: it will increase efficiency, lower the total cost of ownership – the list goes on. Telcos know that they need to move towards an automated and efficient network if they want to support agile service innovation and delivery on a competitive level. But at present, automation has only reached varying levels of maturity across the network span. For many telcos, RAN automation is still limited to discrete trials with small groups of vendors; it’s mostly experimental, and its scalability remains unproven.

Even a few years ago, the complexity of trying to link multiple software systems would have been unthinkable. Today disaggregated RAN is not only possible – it provides a very real competitive edge. The problem is, Open RAN requires a level of intelligent automation that is difficult to build completely in-house¹. Let’s look at the specific challenges to automating Open RAN, and then see what solutions are available.

No one said Open RAN would be simple…

Radio networks are inherently complex. Add to that business requirements that mandate compatibility between new, next-gen networks and legacy technologies, and that complexity multiplies. Operators find themselves facing two options: automation that’s fairly easy to implement, but limited in scope, or automation that links entire networks, but must be custom built, which typically requires some help from outside software experts. Add to that the steadily increasing number of sites², plus the need to keep software expenditures and OPEX in check, and that complexity becomes a serious obstacle.

5G generates a flood of data that – for all the reasons listed above – creates some very real challenges for operators. This data needs to be classified and prioritized for effective network control and management to be possible. The solution? Classic automation is not enough. Open RAN depends on intelligent automation.

Abstracted architecture, concrete benefits

Capgemini solves the issue of multi-vendor CNF diversity through layered, abstracted architecture. In plain English, abstracted means that the architecture is not tied to any specific vendors’ software, but can be quickly tailored to incorporate multiple combinations. It’s the difference between a recipe for a cake, which needs to be followed precisely, and only works for one set of ingredients, and the skill of barbequing. A good grill master can swap out any number of meats or vegetables and adjust the technique slightly, without the need to find (or create) a new recipe each time. In the same way, an abstracted architecture makes it possible to automate a network end-to-end, without the difficulty and cost of a fully unique solution. That goes a long way to managing time, costs and complexity. But the most interesting piece is yet to come.

The spark of intelligence

The heart of Capgemini’s OpenRAN Operations Automation solution lies in a set of RAN applications driven by our NetAnticipate AI-Model platform. To address the issues of automation in real time – when millions of impulses are streaming through networks and each one must be routed correctly and immediately – something more than standard automation is required. The innovative solution we’ve created uses the near Real-Time RIC (nRT-RIC) model. This is based on the extendible, abstracted architecture described above, that enables easy integration of multi-vendor xAPPs on nRT-RIC. So whatever vendors an operator is working with, the same powerful AI is able to handle the traffic. RAN-specific AI models that learn with no supervision, make O-RAN NonRealTime RIC implementation possible. The result is a complex network that essentially runs on autopilot³.

The challenges of Open RAN – the complexity, the constraining brownfield environment, the risk of cost overruns – these all come down to the need to intelligently manage information flow. By doing so, our OpenRAN Operations Automation solution opens the door to a range of benefits.

The benefits of automated Open RAN

Intelligent automation turns Open RAN from a resource-intensive challenge into a source of value. Some of the benefits include:

• The ability to deploy in multi-vendor RAN environments consisting of complex multi-technology networks where the automation platform can create the greatest operational impact.
• Lower operational costs thanks to the automation of network deployment and network operation, leveraging new automation rApps and xApps deployed over O-RAN SMO.
• The ease of harnessing proven operational models of legacy RAN application by modernizing to cloud-native service on nRT and NRT RIC platform.
• The ability to deploy RAN automation across multi-technology networks, using design patterns in alignment with the O-RAN Alliance, and providing future-proof flexible automation across varying technology and vendors. The level of automation can also be adjusted in the network’s constituent layers to varying degrees – for example a high level of automation on RAN compute and connectivity infra, and a medium level in radio resource management.

Looking forward

Solving the riddle of efficient Open RAN automation is only the beginning. With reliable RAN, telcos can create new innovative services like end-to-end network slicing and open and closed loop service assurance, thanks to Capgemini’s RAN automation solution portfolio. With the same underlying infrastructure, telcos can start to capitalize on new business opportunities as service providers for 5G consumer and enterprise services, as well as for mission-critical communications providers. With the right partner and the right solutions, the benefits are just around the corner. Contact me below to learn more.

TelcoInsights is a series of posts about the latest trends and opportunities in the telecommunications industry – powered by a community of global industry experts and thought leaders.

¹ Such automation depends on specialized, real-time cloud network functions (CNF) software management and deployment automation

² Due to network densifications and real-time sensitive Radio Network traffic, which makes it very difficult for any third-party generic automation master controller to seamlessly schedule and operate cloud network functions (CNF), virtual network functions (VNF), or physical network functions (PNF) without a large degree of customization.

³ It’s worth noting that the nRT-RIC does not have a clearly-defined open interface to xApps, so to enable developers from different expertise background to be able to create xAPPs in an Application Builder environment, Cagpemini has introduced a SDK (xAPP SDK) and plugin template approach to xAPP development.  This can easily be integrated over the underlying nRT-RIC.