Can you provide an overview of autonomous networks as a concept?
Telephone networks are large distributed and interconnected systems, composed of many different elements. To ensure that telephone networks enable our phones to access emails, stream videos, and call for help requires hundreds or thousands of skilled engineers to fix problems, perform maintenance, and update technology.
To keep our networks running, without seeing our monthly bills skyrocket, automation of the management of the network is increasingly used. Here, repetitive tasks are replaced by software, freeing skilled engineers to focus on more challenging tasks.
The holy grail at the end of this automation journey is known as autonomous networks. [That is] networks which can monitor, operate, heal, optimise, reconfigure, and govern themselves.
What work has the university been doing in the field, and does the acceptance of the ‘Autonomous Networks - Architecture Framework’ as a standard mean in real terms?
Current approaches have had amazing success in leveraging algorithms, and different types of machine learning, to solve known and well-understood network use cases, enabling cost savings and efficient network operation. This is what we mostly see today. However, this process relies on humans to design and maintain these solutions over time.
Our research focuses on how to move from networks which are automatically performing tasks within pre-defined bounds - level 3 autonomy -, to networks that are able to adapt to the unknown. That is, level 5 autonomy.
Examples of this include ‘how to automatically adapt solutions over time’, ‘how to integrate future technologies’, ‘what to do with new traffic patterns’, ‘how to identify and respond to anomalies’, and ‘can we operate our networks better than engineers?’.
Put simply, we are seeking to automate the process of automation. We are doing this via three key concepts of exploratory evolution, real-time experimentation, and dynamic adaptation. Or create, validate, apply for short.
Working with partners from different sectors, the standard is the first step in describing how this approach can be realised by providing a software architecture based on the three key concepts, and how they practically interoperate. This has been done in a way that enables different stakeholders to bring their own ideas and software, and still be able to interoperate with others.
They can bring their own opinions and implementations on what they need in their networks, but integrate them in a way that follows the key concepts and interfaces of this standardised architecture.
What are the implications for business/mission critical communications?
Autonomous networks present business with a way to not only automate the operations within their networks but to adapt to changing futures, including non-technical business goals. For example, tasking the network to autonomously explore sustainability or low carbon business goals, in addition to maintaining high levels of signal coverage.
Autonomous operation of the network enables faster and customised responses to the live conditions of the network. This aligns well with the new sea of vertical applications and slicing in 5G and AI-native customisations of 6G.
Pragmatically, this means fewer resources will be required for operational aspects of the network, enabling skilled engineers to focus on more innovation and enhancement, rather than maintenance and troubleshooting.
What can businesses do now to take advantage of the opportunities presented by these networks? How will they need to change from an operational perspective?
In addition to the new Autonomous Networks - Architecture Framework, there are many standards activities on different aspects of autonomous networks all of which are positively contributing to the conversation. These include as ETSI’s ZSM, or 3GPP’s SA5.
However, to make autonomous networks a reality network operators need to balance standards development with prototyping and fast failure.
In the Focus Group on Autonomous networks, we have been running an annual multi-month build-a-thon, where challenges based on the key concepts have been set and teams from all sectors and countries have participated. This has given great insight and feedback in the development of the standard, as well as the advantage and opportunity of autonomous networks. Feel free to reach out if you are interested!
Telephone networks have been on a significant journey of digital transformation in recent decades to embrace software. Just as in digital transformation, the largest change that will face business to embrace autonomous networks is a change in mindset. They must learn to stop solving problems and stop creating solutions.
Instead, they must embrace the idea that the autonomous network will create solutions, and their engineers must provide the ‘Lego blocks’ that can be used by the autonomous network to create [these].
Examples of blocks could be new analysis algorithms, interfaces to new network elements, data formats, or knowledge graphs that capture experience. Beyond our own work, this need can also be seen in ETSI’s modular closed loops.
Coupled with this new mindset is the need for trust. While not yet justified, businesses must find ways to gain trust in software-based operation and update of the network.
This has already started today in select areas, but must grow holistically across the network, including the business operational elements.
With the standards increasingly in place, what needs to happen to ensure widespread adoption of the technology by industry/business?
One of the challenges of network operators is that their networks are usually classed as national critical infrastructure, meaning that stability and correct operation are not just desirable, but often legal requirements.
As such, trust is a key element in achieving widespread adoption of autonomous technology. This is important to remember when comparing to other activity in the broader tech sector, where viral technologies are more commonplace.
The research-innovation-standardisation-integration cycle is the right way to move forward with the whole industry. With international digital transformation becoming more and more realised, the ability to program telecommunication networks opens the door to innovation pace of our choosing, and not of technical constraint.
So, to achieve the autonomous revolution is a combination of what telecommunication networks have been doing for over a century. [That is] cautiously and carefully integrating new technologies for the betterment of the network and the business.
The difference this time around is that integration of technologies, such as the Autonomous Networks - Architecture Framework, are empowering telecommunication networks to do the next and subsequent updates autonomously.
