Skip navigation

Ericsson AI: more than a decade of industry leadership

Despite already impacting nearly every industry, Artificial Intelligence (AI) is just beginning to transform the world. With 5G now a commercial reality, the ability to effectively collect, analyze and autonomously act on real-time data is a catalyst towards making the digital economy a reality. Ericsson is making it happen, where AI is nothing new to a company that has been developing and embedding it in their solutions for more than 10 years.  

Mar 26, 2020
Drone deployed over a forest fire
Drone deployed over a forest fire

Across industries, AI is being used in all manner of ways to increase operational efficiencies, lower costs and improve customer experience. Just use a media streaming platform such as Spotify and Netflix and you can see AI in action, as the algorithms analyze your user behavior, and that of people with similar behavior, to recommend new TV shows, films, and music. But these examples, impressive as they are, solve simple use cases with simple AI algorithms.

What is AI doing in more complex industries, with more complex challenges and use cases to solve, where much more is at stake? The financial sector is one that comes to mind, transportation another. But perhaps more complex than both of these: communications. Connecting everyone, connecting everything, everywhere, at any time, on demand, is an enormously complex task, with equally complex infrastructure and technology. AI helps to solve the challenges along the way, but it doesn’t do so with one broad, general brush. There is no one-size-fits-all general AI that will solve all problems.   

Customer focused

Elena Fersman, Director of Artificial Intelligence at Ericsson Research, says the company’s approach to AI is specifically focused on solving the challenges of its customers and the telecommunications industry for many years.

“Our AI does not power talking cylinders, create art experiments, label photos or recommend movies and music. It is not a general-purpose tool intended for any use case imaginable, and we don’t have brand teams think of cute names for it…we aren’t interested in any of that,” Fersman says. “At Ericsson, we have been working with AI for more than a decade and take a different approach. We develop AI to solve the right problems: the right problems for our customers, and not problems outside of those. We are embedding AI deeply throughout our portfolio and services specific to use cases that can scale to all our customers.”

“Because of this approach, we don’t sell or market AI as a solution in itself. It comes with what we offer, as standard. I think, perhaps, this has led to a perception that in some way we may not have been associated with AI in the same way as companies who focus only on AI. But for the specific needs that we focus on, our AI expertise and leadership is unsurpassed. We have proven focused industry expertise in AI, so in that sense we are an AI company.”

Six challenges

It’s with this laser focus on solving the right problems, not all problems, that Ericsson has identified six specific challenges for communications service providers that, when addressed with AI-embedded solutions, add significant value:

  • Ensure network performance
  • Improve energy efficiency and meet sustainability demands
  • Turn operations into an efficient business enabler
  • Realize the value from cloud and a virtualized core
  • Create an outstanding, future-proof customer (user) experience
  • Bring out the most from existing infrastructure

“Focusing on these six specific challenges for our customers allows us to be very concentrated when developing and embedding AI solutions that truly add value,” Fersman says. “For example, to improve energy efficiency and meet demands of sustainability, we have developed and embedded an AI solution for automated MIMO sleep in radio base stations that provides 14 percent energy savings when turned on.”

“When we focus on turning operations into an efficient business enabler, one solution within our Managed Services Offering – the Ericsson Operations Engine – is able to predict KPI degradation in the Radio Access Network, and now more than 70 percent of KPI degradation incidents are prevented with data-driven operations.”

“The fact is, for each one of these six specific challenges, we have multiple live AI solutions operating in live networks for our customers that are already adding real value, whether through minimizing costs, reducing incidents, increasing efficiencies, curbing energy consumption, or improving user experience. The list goes on and it is something that I am very proud about.”

The future

So, what does the future of AI look like at Ericsson? Fersman says it’s about solving complex problems now, and a broadening AI mind in the future. “The more complex problems that we can solve now with real and tangible value for our customers, the more that will add to a broader AI implementation across our portfolio as we increasingly make these AI-powered solutions work holistically across the network,” Fersman says. “Imagine networks that run themselves, optimize themselves, solve problems autonomously. That’s where we are heading, and we are going there with speed.”

Ericsson: AI by design
Ericsson Operations Engine
Ericsson gated report: Supercharging customer experience through AI and automation


Q: What sets Ericsson apart in the telecommunications industry with Artificial Intelligence?

Fersman: “We have five main Ericsson AI differentiators that truly set us apart from the competition: Telecom & AI expertise; AI-ready technology; AI use case co-creation; unique access to data; and an AI-powered portfolio.”

Q: Let’s drill down into each of these five starting with Ericsson’s Telecom & AI expertise.

Fersman: “Ericsson AI is built on a strong foundation of combining deep telecom domain expertise, data science and AI knowledge. These skills always work hand-in-hand, close to customer problems, and collaborating to solve them. This means that we know what the right problems to solve are, how to solve them and what a good solution looks like.”

Q: And what about AI-ready technology?

Fersman: “Ericsson’s technology is AI-ready – from RAN, to centralized clouds, core and operations – to customer-facing areas such as BSS and customer analytics. We have made architectural decisions that make AI easy to deploy holistically across the network. Combined with our telecom domain expertise, this means we can use AI where it makes sense, solving the right problems for our customers.”

Q: Tell us more about Ericsson’s approach to AI use case co-creation.

Fersman: “We have learned through experience that co-creation – developing solutions together across competence areas with our customers – is critical. We develop our AI in the field together with our customers with real network data, leveraging our domain expertise and a portfolio of ready-made use cases and pre-trained algorithms that we adapt to various customer contexts and problems. This way we can solve the right problems for our customers, instead of providing an untrained general platform or solutions that only solve very specific challenges.”

Q: What access to data does Ericsson have that is unique in terms of AI development for the telecommunications industry?

Fersman: “We have a deeper level of access to data than communications service providers for training our algorithms. Since we support and monitor our networks, we get access to very valuable data that is perfectly suitable for AI development, and when complemented with unique access to a wide variety of operational data through our Managed Services offering, we can develop algorithms that are more powerful than those offered by other players in the industry.”

Q: Please tell us more about Ericsson’s AI-powered portfolio.

Fersman: “We are not a niche player, focusing on one use case. Instead, we build AI into every part of our portfolio where it makes sense, all the way from baseband to centralized clouds. With AI spread out through all of our products, we can see the whole end-to-end context and how we make every part of the network function together. Our portfolio also comes with AI embedded – it is built straight into our products and ready to be turned on.”