How AI and Ericsson experts are revolutionizing service continuity in networks

As with many of Ericsson’s great innovations, the story begins with a simple request from a customer looking for assistance. Back in 2017 a client in Asia was planning a wide-ranging expansion that was going to expand the size of their 4G RAN network.

In the planning phase they reached out to Ericsson and said, “we don't want to have to double or triple our network operations resources when we do this expansion – can you help us to automate as much as we can?”

We got our customer support team and put them together with some off-the-shelf AI apps, along with an innovative new set of automations and use cases - these are the first versions of what eventually became known as AI Apps.

Instant success – and a new way of working

Eventually that client’s network grew by a factor of seven, without having to dedicate more resources to operations and maintenance. The client was of course delighted with that outcome, and as a result the bar was raised – if we could do this for a 4G expansion, what else could we automate to help them?

The next big challenge they wanted us to address was fault isolation on their IP Multimedia Subsystem (IMS) network. At the time they were using IMS software that was quite old, and pretty much everything was being done manually with very little automation.

Our customer services team got to work once again, and in a short space of time they were able to produce an automation that helped them with fault isolation that reduced the time from around 30 minutes to seven minutes – and not only that, but the automation also increased accuracy.

Almost without realizing it, we had created a new way of working together with this customer that would create a template for how we approached service continuity issues in the future – once a year, or more often if needed, we would sit down with the customer and discuss their business priorities and their pain points. Together we would agree on a number of focus areas for service continuity, and then we would go back to our experts at customer support.

From there we would create AI apps to address these pain points – sometimes they were existing AI apps that we had built up in our library, and sometimes they were newly developed to address their specific needs.

How an AI app led to a 43 percent saving on hardware

To show the effect these AI apps have had on the bottom line for our customers, let’s look at one very specific case related to a specific term used by customers in service continuity – “no fault found”.

This was an internal description used by the customer when they got an alarm that indicated that some piece of hardware in the network was faulty. The customer experience is the most important thing for them, and their network can never be down, so as soon as they get one of these alarms, they don't even bother troubleshooting because that takes too much time - they just swap out the hardware straightaway, rather than risk a network outage.

This approach is, or was, very common, even though it may not necessarily have been a hardware fault, and it might even have been a false alarm, but CSPs would prefer to take the expense of swapping out a piece of machinery than risk the network going down.

That meant that this customer was replacing a lot of hardware that wasn’t necessarily faulty, which was costing them money.

Together with the customer we were once again able to create an AI app, and the difference it made was enormous – we managed to reduce their expenditure on “no fault found” cases by 43 percent. By automating the fault-finding, we could be much more precise, which meant less new hardware and less site visits.

What is an AI App?

Though many CSPs still have legacy technologies in their networks, as a rule they are keen to try the very latest technologies, provided of course that they are going to work in their environment. There is a huge desire among CSPs to embrace things like AI and machine learning (ML) and operators are very much aware of both the benefits of automation and the potential for savings when it comes to service continuity.

The AI Apps that make up the Ericsson Service Continuity App Suite are basically intelligent algorithms designed for a specific purpose and with a specific outcome in mind. They are not very generic in that they address very specific issues; that could be addressing a silent issue, it could be addressing hardware and how it is functioning, or it might address an area such as energy consumption, which is very much a hot topic with customers.

AI app Suite Tech Unveiled episode

The idea for service continuity is that there should be a suite of these algorithms running in the customer network, and they should look for anomalies and things that are not right. They use a logic that says, “this doesn’t look right”, and either report back or take immediate remedial action without any human interaction whatsoever.

In looking at these anomalies, the algorithm learns about trends in the network and that enables it to address issues before they become a problem, making the network run smoother and continuously improving energy consumption.

Even though each app in the AI App Suite addresses a very specific issue and is thus not generic, they are mostly highly configurable, meaning that they can be adjusted to work from network to network and to operate in different ways, depending on what the network operator wants them to do.

Powerful customization

Fast-forward to today and for the most part the apps are built in the same way using an AI App template. We have templates for different issues like fault isolation, energy efficiency or latency, and then when we go to a customer, we customize that template for their network.

Because everything is already set up on the template - all the machine learning, all the complicated stuff - we just need to do some fine-tuning, and then it works for that specific customer to address their specific problems.

That makes it very easy for our customers – they don't necessarily have to have a team of data scientists to develop a use case because we already have these templates that we can customize for them, and we can work together to customize whatever they need quite quickly.

The value of AI apps in service continuity

When a customer buys hardware from Ericsson, they can essentially be guaranteed that they will get access to all the features and functionality the moment they take it out of the box, but the same cannot be said for services like service continuity – what can be said, however, is that there is huge value to be had from investing in such services, even if it does differ from customer to customer.

As indicated above, many of the AI Apps in the suite have their genesis in a specific customer issue that has been highlighted to us – while we have created several of our own in our lab environment, the vast majority come from real-life situations experienced by our customers, and many in the team would say that collaborating with them is the most satisfying part of their work.

But not every customer has the same needs, and an AI app that might be indispensable for one customer might be of no use whatsoever to a different one. What we do see, however, is a great degree of reusability from customer to customer – the network environment might be the same and the AI and ML elements will produce different results for that reason, but the core issue will remain the same, or at least have some similarities. Thus, discoveries made with one customer can often have relevance to another.

Once the apps are up and running and the AI and ML elements take over, processing the data and learning about the specific network, it can be even more difficult to see the value they bring, but rest assured it is there. The App Suite can be compared to an insurance policy, in that you often don’t realize you have it until you really need it – until then it works away in the background, ensuring that everything keeps working as it should and preventing things from going wrong long before they occur.

The long-term value of AI and ML in service continuity

Communication Service Provider’s needs will evolve, and our customers can adapt their portfolio of apps and the AI apps themselves will evolve as new data input is added, algorithms are configurated and their capacity to adopt the AI apps matures. The AI apps are also flexible and can be deployed to different customer environments and infrastructures.

Though, AI Apps will remain a key part of Ericsson’s service continuity offering – there is a constant desire for innovation from customers as they discover both the power of existing AI apps and the potential for future apps.

AI apps tend to have a very quick time to market – once the issue is identified together with a customer, a template is chosen, and the work begins on finding a solution to that specific problem. There have been instances where we have been asked to address very specific issues around purpose-built networks. For one major sporting event we came up with an app in just eight days to address an issue that the organizers were having. Needless to say, this kind of speed and flexibility is something that customers find very useful and valuable.

It’s important to note the difference between Ericsson producing AI apps in our lab or testing environment and those that we work together with customers on. When our experts work on these things on their own initiative, they often do so based on what they perceive to be a problem in the industry; when we work together with customers on them, we know that the problem exists in the real world, because the customer has told us so, and thus the results of such co-operation are often of greater value.

The vision of the future

From its humble beginnings based on a question from a customer, Ericsson’s AI App Suite for Service Continuity has become almost indispensable for those operators who have deployed it, and more and more are realizing the benefits that the app suite can bring to their business.

In our vision of the future of service continuity, we see a situation where CSPs have integrated the AI app suite as part of their normal operations and daily activities - when the customer gets into the office in the morning, the first thing they do is to look at our portal to see what their priorities for the coming day should be.

The goal is to help and guide them to continuously improve network performance – just by glancing at the portal they should be able to decide what they need to work on that day, what site visits to prioritize, and so on.

In time, CSPs will realize that they are sitting on a very valuable asset that is also indispensable both to their operations and their plans for the future – the vast amounts of data that their networks constantly generate.

Analyzing this data in real time enables them to make much more informed decisions about what they are doing day-to-day, and how these things can be improved in the future. With data about everything from traffic and energy efficiency to security, they can have the courage to innovate, knowing that they can reduce, and in some cases completely eliminate, the risk that comes with upgrades and changes in their network environment.

The origin story of AI apps is one of the most fascinating things we have seen in the industry in recent years, and the next chapter promises to be even more interesting – and more and more operators want to be part of the story.

Related content

Why is the interaction between AI and human guidance indispensable?

What does my network do when I don’t look at it?

Learn how Service Continuity AI app suite supported a sustainable, healthy network and saved 25% energy in a collaboration with Far EasTone