• The user dream could become a nightmare for vendors if AI can realise the vision of decoupled hardware and software 
• But many challenges remain, notes the University of Bristol’s Professor Dimitra Simeonidou in the second part of her interview with TelecomTV
• It is vitally important that clear rules and strong guardrails be implemented for the use of AI in networks
• How will the telecom industry keep up with the pace of change as the traditional ‘standardisation cycle’ is increasingly disrupted?

Many big names are currently prophesying major changes in the future of telecom, usually in ways that would benefit themselves and their shareholders. 

Professor Dimitra Simeonidou of the University of Bristol, meanwhile, has spent the past 30 years working on ways to improve communications networks for the common good and her current focus is AI-native networking and its impact on high-performance, programmable infrastructure. 

She recently talked at length with TelecomTV about how she believes the telecom sector is evolving and what needs to be done to develop the very best next-generation networks – you can read the first part of her interview and find out more about her background and achievements, in Dimitra Simeonidou: Embedding AI into telecom networks.  

Simeonidou is convinced that, as the epoch of ‘end-to-end AI’ dawns, artificial intelligence must be considered a vital, integral part of the network and be integrated in the foundations of networks. But embedding AI in telecom will require a radical redesign of communications networking architectures. 

Simeonidou is keen to address the most important issues that network architects now have to address. “There’s still a lot to learn around AI networks as critical infrastructure,” she notes. Many aspects remain to be settled, including, at the fundamental level, how to ensure that AI can be used safely, where cloudified AI can be allowed to run autonomously, and where (and when) human oversight of an AI network will be absolutely vital.

Evidently, AI offers an enormous opportunity, but “we need clear rules and guardrails around the use of AI in networks,” notes the professor. The industry needs time to learn how to use AI effectively and, in due course, to reach the point of “supporting open networking and increasing diversity of the supply chain.” 

She adds: “I’m not just talking about the RAN here but the whole of the network itself. AI is an opportunity to finally make tech infrastructure agnostic,” Simeonidou boldly proclaims. 

Then there is the ‘white-box paradigm’ which, in the context of comms networking, refers to the decoupling of network hardware from network operating systems. Thus, instead of purchasing proprietary, bundled hardware and software packages from a single vendor, organisations and businesses can turn to standardised, off-the-shelf ‘bare metal’ hardware and independently install their own choice of open-source or third-party software. As Simeonidou says, “We now have intelligence that can run over hardware and software to be able to interoperate across different technologies and make everything pluggable and replaceable.”

She adds: “It has long been a dream to balance the power of the equipment vendors” and see the creation of an “open diversified environment where we can [introduce] new innovations so that service providers can no longer claim, ‘Oh, we can’t put this new vendor equipment in our infrastructure’.” She believes “there is a real opportunity here. There are counter arguments for both sides, but there is an opportunity to save the industry.”

Technology agnostic networking – an opportunity and a threat

Another opportunity identified by Simeonidou is that “networks are increasingly capable of reading machine code… most new standards come in machine readable formats and can apply [updates] across the network… you have programmable hardware and tools that allow programming at network scale”. In that scenario, “the identity of a vendor no longer matters.” In more ways than one, AI is providing vendors with opportunities but also poses a threat if it can enable programmable telecom infrastructure at scale in a way and at a pace that the vendors cannot control.  

But there are also many knotty problems associated with the standardisation and regulation that are both critical but can also hinder progress.  

Simeonidou is adamant that regulation is vital to drive the industry and the technology forward. “We must have a powerful regulatory framework and regulatory discussion between all interested parties working together to help format an overarching policy, especially in light of the emerging need to share spectrum” in a wireless world that needs cellular and Wi-Fi access: She stresses the need for all parties to discuss matters in advance of regulatory intervention being enacted.

The professor also believes that within 10 years, AI will prevail across almost all of the world’s telecom infrastructure and, alongside that, a new and very different standardisation environment will have to emerge as network transformation inevitably takes place. The reality is that change is happening at an accelerating pace and, as usual, is well ahead of the establishment of new standards (6G being a current case in point). The big question is “how will the industry cope with the incredible rate of change, especially as the ‘standardisation cycle’ is increasingly disrupted?” 

As Simeonidou notes, standardisation always lags behind the realities of technological advancement and the pressing need to balance the speed of the proliferation of AI is of such importance that “it has to be considered globally”. She wonders, “What will happen to standards? Who is going to wait for seven years for a standardisation cycle to run its course when the likes of Musk, AWS, Google and other hyperscalers already provide services?”

Finally, whilst Simeonidou has direct, personal, practical experience of spinning private companies out of academic research, she points out that such a strategy is not the only route to translate university research and experimentation into commercial enterprise: She cites local engagement and the development of innovation pipelines and community services as increasingly popular and valid alternatives.

– Martyn Warwick, Editor in Chief, TelecomTV