By: Shamik Mishra

5G is finally here. Well, almost. After all the hype, operators are slowly but surely deploying the technology. And they are turning to automation to address the demands placed on the network from the explosive growth of data and devices. The ability to scale services quickly and flexibly is the key to success—and network automation provides operators that capability.

That is easier said than done. All too often, operators have ‘reactive automation’—a rules-based system that springs into action after a failure. In most instances, subscribers can feel an impact on the quality of experience. That is not really fit for purpose in the age of 5G. Instead, it really should be ‘proactive automation.’ That is when artificial intelligence (AI) and machine learning (ML) algorithms are trained to take corrective action before a failure occurs by predicting network behavior. Such automation is designed to negate the possibility of a subscriber experiencing a service shortfall.

Many reasons to pursue automation

A growing number of operators recognize that the network efficiency brought about by automation is integral to their ability to manage the complexity of 5G. This is especially true with heterogeneous equipment meant to deliver end-to-end services. An early example is Vodafone, which increased its network optimization speeds by a staggering 45,000 percent by implementing AI-enabled augmented engineering.

Augmented engineering involves the effective use and analysis of diverse data sets that are usually executed with specific computing software. This can be time-consuming. According to Vodafone, machine learning made the process efficient, so engineers focused their attention on more critical and strategic projects while the automation did the heavy lifting. Research from MIT Technology Review shows that operators expect automation will result in operating expense reductions of between 30 and 50 percent.

Futuriom, a technology research company, conducted a survey of over 200 operators that revealed their top three goals for implementing network automation. These goals included:

1. Faster service delivery to increase revenue (79 percent);
2. Boosting network security (75 percent); and
3. Improved ability to support dynamic and on-demand services (70 percent).

Interestingly, respondents considered lowering operating costs (42 percent) and capital spending (38 percent) as secondary goals. Moreover, when asked where they wanted to implement cost savings over the next two years, 63 percent said they want the ability to use intelligence and analytics to automate fault resolution. Fifty-five percent wanted the ability to automate network design, build and deploy operations.

One size does not fit all

Simply put, there is no one-size-fits-all path that leads to network automation. Some operators start by automating in the data center. Others focus on automating operational support systems (OSS). In most cases, operators generally begin by automating virtualized functions (VFs) before tackling legacy systems. This variety shows that there is no right or wrong way. There are a number of emerging use cases that suggest how network automation could shape the network of the future.

Mirroring the findings of the Futuriom survey, Altran’s extensive experience with operators has found that automation is the top choice to prevent faults. Consider these other use cases:

• Security: Automating the network operations center can prevent distributed denial-of-service (DDoS) attacks and botnets. This strategy enables root-cause analytics, anomaly detection and guided diagnostics to strengthen security and address vulnerabilities at the onset.

• Network planning: In the era of unpredictable traffic patterns, automated intelligent service rollouts can manage the sudden peaks and troughs on radio and fiber optics to deliver consistent quality of experience. 5G infrastructure will be reliant on cloud-native architectures and most of the network function will be “workloads.” Automation will need to consider “compute planning” as well as network planning.

• Network slicing: Automation has also given way to intent-based dynamic network slicing. Thanks to this capability, operators can secure new enterprise revenue streams by offering bespoke network slices to meet different use cases and business models.