You wouldn’t expect a conference focussing on ultra-reliable, low-latency techniques, technologies and requirements to be anything other than highly technical. And the recent URLL 2017 meeting in London didn’t disappoint. The air was heavy with acronyms, as engineers sought to reconcile technical ambitions, and practical and physical constraints. The pleasant change was the grappling with the commercial realities of realising the requirements.
However, in the context of 5G – which, for many, is more about connecting things than people – the theme of the conference was important and the issues not that difficult to understand.
With today’s mobile networks, customers (in this context, enterprises) mostly get what they’re given – and do their best to make it work for whatever they’re trying to accomplish. But 5G isn’t meant to be like that. One of the many things it’s meant to be about is the ability to deliver platforms through which customers can select what they need in terms functional architecture and SLAs. And, of course, pay a significant premium for the privilege of so doing.
In this way, the argument goes, the new networks will be flexible and robust enough to support a wide range of industrial automation: massive machine-type communication (MTC) – both low-rent MTC, that connects lots of low-cost and low-energy devices, and ‘mission-critical’ MTC, that delivers real-time control and automation of dynamic processes.
It’s the mission-critical services that need to be ‘ultra-reliable and low latency’. So, what do we mean by ‘reliable’? Well, it depends who you’re talking to, but in this context reliability is framed in terms of delivering messages successfully within a determined latency ‘budget’ – and, at the top end, according to ETSI, only one message in one billion data transfers may be lost or delayed. Latency is simply the delay between data being generated and correctly received – which, according to the ITU, can be as little as 1 ms end-to-end for tactile Internet applications. Requirements such as those from the ITU set a level of expectation for the research and standards setting communities, but do they really meet the needs of the systems in the verticals?
I came away from the London conference with an even longer list of ‘issues that need addressing’ than I went in with. There’s clearly a huge amount of technical detail round architecture, frame structure, packet size, channel uses, block error rates, diversity and re-transmission techniques that need to be sorted through use case assessment, evaluation and standards. Network slicing has a significant role to play in all this, but the standardisation of slices and their orchestration continue to develop. And, as quite a few speakers pointed out, reliability and latency are actually two very different challenges that, while related, shouldn’t be conflated.
But for me there are even more fundamental issues that are likely to stand between ambition and delivery.
The first is about really understanding end customer service performance requirements. I think this is where Real Wireless has an important role to play, helping businesses communicate real world requirements of the industrial systems and automation URLLC is meant to address. Today, the mobile industry is very much focussed on designing and delivering on its own terms in its own way. Real Wireless is well placed to act as a mediator or innovation broker, helping each side understand the language and priorities of the other. The jury is still out on whether a MNO use case driven infrastructure can effectively address vertical driven use cases that are established to capture value in a vertical industry value chain. Would a better solution be for the vertical to build their own infrastructure, applying neutral host principles to assure connectivity quality?
The other big change that needs to happen is cultural: establishing and – importantly – maintaining a service. Service providers are used to delivering a ‘you get what we get’ offering to enterprise, with bespoke service delivery very niche and extremely expensive. Slicing starts to address the reframing of the business relationship and creating “templates” around systems and business models that are absolutely focussed on the very specific needs of a very wide range of enterprise demands. The emergence of options to automate the establishing of user and tenant relations helps; however, there is no doubt that very different approaches to client management will be needed in this B2B world. This is generally not something a majority of mobile operators are well set to deliver. Once again, Real Wireless is well placed to support both sides in what seems likely to be quite a steep learning curve.