Communication Service Providers (CSPs) are under increasing pressure to reduce both operating and capital costs as 5G network rollouts continue to gain momentum. Artificial intelligence (AI) and machine learning (ML) are quickly emerging as powerful tools to help operators do just that by aiding in complete network management.
And while CSPs have historically been reluctant to put their data on public clouds, running AI algorithms for network optimization over a public cloud offers several key benefits and protects data thanks to Nokia’s security framework for public cloud. Therefore, according to Nokia’s Head of Cognitive Services Ajay Singh, it might be time for operators to rethink their stance on both AI and the public cloud.
As 5G, and the applications it enables, adds complexity to existing networks, the use of AI to manage and optimize these networks is becoming a requirement.
“It’s not only 5G adding to the complexity to the network,” Singh explained. “You now have many different layers of cellular technology on top of each other when you consider 2G, 3G, 4G, and then the addition of 5G. Then the new use cases and abilities, like network slicing, come along.”
This, he continued, makes it “impossible” for humans to efficiently operate modern networks. Without AI, operators would have to significantly increase the number of network engineers and field operators in an attempt to keep up, a move that would drastically increase operating costs.
However, building in-house solutions would require considerable time and financial investment, and despite the clear need for AI-based network solutions, a study of 50 CSPs showed that 56% of CSPs face data quality issues, 77% struggle with data storage and consolidation and only 35% of those who aspire to lead in the field of AI believe themselves to have “advanced” data capture processes. These facts suggest that CSPs need to partner with AI solution providers to help them reduce up-front investments and risk, while ensuring the highest possible level of data security and subscriber experience.
An AI-as-a-service approach can be a great way to ensure flexibility because these models are consumption based, and therefore, operators only pay for what they consume and, in many cases, are not required to pay a fee if they do not see positive results.
“The as-a-service model allows for tailor-made solutions and is supported by AI algorithms that are trained based on global data that delivers insights into the real challenges that CSPs face,” explained Singh.
These algorithms are so powerful that they can result in a zero-touch network, or a network in which fault detection and resolution occur as closed loop actions, meaning that the resolution is immediately addressed by the machines themselves.
Because of the hesitancy CSPs often feel about new services, they typically require proof of concepts (PoCs), the implementation of which cost a lot of time, money and effort. However, delivering AI solutions over a public cloud reduces all three of those pain points because a shorter time between PoC and live deployment can be achieved. Doing so also allows CSPs to ramp-up and ramp-down cloud compute, storage and connectivity resources as required
Further, running AI solutions over a public cloud offers much more flexibility and scalability than doing so on a private cloud, allowing companies like Nokia to help a customer scale up a solution from network cluster to nationwide level in a single day. Network operations are also conducted faster over a public cloud and new solutions, or network features can be quickly tested without impacting the overall network KPIs.
“With the public cloud, network solution deployment is as much as nine times faster,” said Singh. “Once the data pipeline is established, we can add new solutions on top of the existing ones within a week. It’s flexible, easy and agile.”
To address operators’ security concerns, cloud companies like Microsoft Azure have security features built into their public clouds, but for an added layer of assurance, operators should work with solution providers that build additional security protocols and frameworks into their solutions that sit on top of the ones established by the cloud provider.
“At Nokia, our added security framework ensures that each customers’ data is isolated, and before it’s sent out to the cloud, it’s scrambled so that it adheres to zero trust protocols,” stated Singh.
By implementing AI and ML solutions that apply global learnings and data-driven techniques, operators can automate their network operations and service assurance, cut costs, increase agility and boost subscriber experience, while their engineers can spend more time solving creative problems instead of menial ones.
Hear more from Nokia and Microsoft experts about the world’s first commercial deployment of the AVA AI-as-a-service use cases library on the Microsoft Azure public cloud platform, watch Nokia & Microsoft webinar.
