Skip to main content

AI and Future of Telco 3 of 3: Automation is foundational - and also really, really cool

Do we need network automation to make the telco sector “viable”?

Automation is foundational - and also really, really cool
Written by
Guy Redmill
Published on
29 Sep 2023

Do we need network automation to make the telco sector “viable”?

According to no less an authority than the European Telecommunications Standards Institute, the answer is yes.

Not just yes—“urgently” yes.

Why? Because we need to automate to hope to cope with all the myriad operational and business challenges 5G and 6G will open up, ESTI says. These centre on the need for the “unprecedented” operational agility we will need to allow services to be rapidly deployed dynamically adapted, for example. At the same time, we’re going to need really good tools to ensure network performance, full coverage, that network capacity will be able to satisfy the requirements of active services, and quickly help us find fault or degradation that might impact the services. 

The thesis is that we need automation because it’s well beyond what human teams could deliver, but also to do it in a way that is economically sustainable and can support this highly diverse service portfolio. Step forward the AN—the autonomous network of tomorrow.

It’s hard to argue that something like this is going to have to happen. 5G technology relies on network automation for several reasons: To take just one example—if you will a ‘slice,’ ho ho—of the task here, to make network slicing anything approaching viable as a business area, you’d have to have a) an easy, repeatable way of creating >1 virtual networks on a shared physical infrastructure b) work to make sure each network slice could be quickly tailored to specific applications or user requirements to win business from paying customers who can’t see an alternative to using it and then c) somehow economically orchestrating and managing these slices as their operator, ensuring the right levels of resource allocation and dynamic provisioning was happening, based on (ideally) real-time demand.

Achieving telco network automation will be challenging

You could do that with an abacus and an infinite set of monkeys… or, you could get AN to help optimise and automate as much of it as possible to keep yourself sane, and able to do other things in your working day.

We can’t say we want 6G to be scalable, complex, and better than what users have now unless we put in the right infrastructure to make it run right and in a way that we can make money out of by it being a defined product that we can let run in the background. 

At the same time, ETSI and many others will also tell you that achieving telco network automation is a challenging thing; the TM Forum said as recently as January 2022 that while ANs are accepted are the key to making the concept of zero-touch operations a reality, CSPs’ progress towards full AN maturity is lagging.

100% true… but we are also, uniquely when compared to our colleagues and - perhaps increasingly - our competitors in enterprise IT, an industry based on agreeing baseline standards that help everyone—then keep improving those standards to encourage growth that raises all boats. 

The telco ecosystem is also full of people also not frightened of a bit of hard work either, so we don’t see any need for pessimism around standards coming along in the nick of time to save us. In fact, their broad outline can already be discerned. ETSI’s ‘ZSM’ (Zero Touch Network and Service Management) group says 2022 saw some major progress on the network automation transformation journey, for instance, with four major specifications and reports getting published on everything from cross-domain, end-to-end (E2E) services lifecycle management to some really interesting work around intent-driven autonomous networks (“Intents declaratively express all the operational expectations an autonomous management domain needs to fulfil and assure, including requirements, goals, and constraints”). 

The group also conducted what seem like useful exchanges of information with groups like the O-RAN Alliance (and again, the TM Forum)—moves which analysts have welcomed (“Collaboration among SDOs is critical to accelerate telco automation, and ETSI ZSM is the glue that holds all of them together from an end-to-end automation perspective”—Ali Rao, Analysys Mason).

Vital as ZSM’s work is, it’s not the only game in town—which is fine if you have a LOT to work on, as 5G network automation involves. Thus the International Telecommunication Union has to be given kudos for the progress of its Focus Group on Autonomous Networks (FG-AN) (and, in fact, this is another one of the research groups ZSM is in ongoing dialogue with): we note with interest that its specific brief is to draft technical reports and specifications for autonomous networks, including exploratory evolution in future networks, real-time responsive experimentation, dynamic adaptation to future environments, technologies, and use cases.

‘Artificial Intelligence will be the key that unlocks the door to Autonomous Networks’

Moreover, the Group will also identify relevant gaps in the standardisation of autonomous networks, and has already published some high-level thinking about likely AN use cases.

For sure, it is difficult to predict when fully autonomous telecommunication networks will be a dependable technology, as it depends on various factors, including doing very clever things we don’t know how to do quite yet.

It’s almost certainly the case, too, that Artificial Intelligence will be the key that unlocks the door to Autonomous Networks, and we’re seeing insanely fast progress here of course (though hopefully not all of it just in helping lazy people write documentation). 

Get this right, and the best technical and agreed AN standards will be the foundation of a great telco future—and as we’ve argued in the previous two blogs in this series, standards will also sort the mess AI currently is in regarding trustworthiness, too.

In this series, we’ve also note that we don't have much time to de-risk AI, and so our help could be of great value there as well—so let’s repeat that message one last time:

To make 6G work, we need AI to work.

And AI only works if we can help it get to the right standards basis—it’s absolutely in our market’s interest to do so – and future success hinges on it.

So… let’s crack on and make it happen.

Get in touch

Need more information on what we do? Enter your email address below and we’ll get back to you.

From the blog
The latest industry news, interviews, technologies, and resources

Repurposing of what we already have installed around us makes both economic and Net Zero sense

Back in May, BT Group announced it had changed a former cabinet unit in East Lothian, Scotland—as in, a cab...
11 July 2024

The advanced telco tech research quandary, Part 2: Do you want to be Steve Jobs… or just a tyre kicker?

In Part 1 of this discussion, we reviewed—deliberately sceptically—some of the congratulatory tone of the T...
28 May 2024