Last year, the Department for Digital, Culture, Media and Sport (DCMS) awarded £16m to the Universities of Surrey, Bristol and King’s College London to fund the creation of a 5G test network, to be comprised of three small-scale mobile networks. Over the next few pages, we’ll get a feel for what each of these research groups has been up to, their areas of focus and the use-cases they’ve been exploring.
Our story starts with a trip to the University of Surrey’s 5G Innovation Centre (5GIC). Professor Rahim Tafazolli, director of the 5GIC, and Stuart Revell, managing director of RTACS Ltd, which is working with the 5GIC on definition, external engagement, and 5G and IoT adoption, are on hand to show me around.
Tafazolli wastes no time, quickly ushering me into a room (shown overleaf) where a two-wheeled robot is performing a delicate balancing act made only possible by a feedback/control loop run between a cloud-based edged computing application, a 5G base station on a pole outside the building running at 3.5GHz, and a prototype 5G user equipment device connecting the robot to the network.
It’s an illustration of the ultra-reliable low-latency aspect of 5G, which will be fully standardised in 3GPP Release 16. The total round-trip delay is just four milliseconds, which includes the uplink and downlink connection and the edge computing time; and this level of service is guaranteed to be delivered 99.999 per cent of the time.
I give the robot several shoves, but my efforts can’t do much in the face of its self-righting behaviour. The same can’t be said for when it’s switched over to 4G, at which point it has all the balance of a concussed lemming and has to be carefully held in place to prevent it from crashing to the floor. The end goal of this technology is to be able to reduce the cost of robotics by putting all the intelligence on the cloud (see last issue’s insight piece on manufacturing). The same technology was used in November to demonstrate remote driving, with a driver at the London ExCeL centre controlling a vehicle on the University of Surrey campus moving at 20mph.
The 5GIC’s wider test bed consists of 44 4G small cells, three macro cells and three 5G base stations, soon to rise to 12. Part of the thinking behind this is to be able to get a feel for how 5G will behave in a high-interference, high-capacity environment – “we want to break it to find out where the problems are and solve them,” says Revell.
I’m then shown the 5GIC’s virtualisation test bed – a large server room (shown above) – which runs the centre’s novel 5G Flat Distributed Cloud Core, which Tafazolli claims is the first virtualised mobile network, has 10 x 10Gbps of fibre connectivity, and can serve up to a million users. The 5GIC’s members, 26 large organisations and a further 55 SMEs, can use it to set up a test of different use-cases and technologies of their own remotely. “On our architecture,” Tafazolli says, “we can set up a network slice from scratch with no previous knowledge in less than two minutes, and collapse it in the same time.” He adds that the 5GIC is looking to put this solution forward to 3GPP to inform the standardisation of network slicing in Release 16. “To our knowledge, this is the most advanced, dynamic and vendor-neutral way of doing network slicing.”
Revell adds that if a company were to request a network slice from an MNO or a fixed operator today, from a technical perspective it could take around nine months to set up, with a further three months for the creation of a service-level agreement and the other contractual arrangements. “This is where you need software-defined networks and virtualisation techniques, but the point where every asset in your network is virtualised is a long way off,” he says.
We move into a small laboratory, where a 5G small cell is being developed, complete with extremely complex schematics on a whiteboard, a massive MIMO 3.5GHz radio head with 128 antenna elements, and another one with 64 antenna elements. I’m also shown the IoT Egg, an IoT device that has Wi-Fi and BLE modules, together with vibration, dust, particulate, temperature, humidity and gesture recognition sensors; this provides a comprehensive small form factor reference design that could be repurposed for multiple applications, such as in an office-like environment to improve well-being.
Revell says one headache for those seeking to capitalise on the opportunities presented by 5G is that doing so requires people with skills across three disciplines – IT, communications and control systems – and “it’s very rare that people have skills across those three domains; this problem is not unique to the UK.”
Similarly, Tafazolli notes there is a big difference in outlook between those in the IT world and their telecoms network counterparts: the former “normally say that if it works, it’s alright; there’s not a good understanding of quality of service, reliability, availability”.
Because of these issues, the University of Surrey has set up a laboratory next door to the 5GIC called Innovation for Health, which as its name suggests is aimed at providing medical undergraduates with the knowledge they need to be confident in using digital technology in their future careers.
Tafazolli highlights the 5GIC’s influence on 5G development. “If it wasn’t for the 5GIC and its industrial partners, 5G would have ended up 10 times faster than 4G; that’s it – full stop.” He adds this would have been a wasted opportunity because it has been possible to achieve this speed improvement with LTE, without any need for a new standard. Similarly, he says the 5GIC and its partners “were quite instrumental” in getting the three pioneer frequency bands for 5G – 700MHz, 3.6GHz and 26GHz – agreed at the European level.
One of the less-understood aspects of standards development is its implications for those who contribute to it. Tafazolli says should the 5GIC and its partners’ vision for 5G be adopted to a large extent in 3GPP, it would be easier for the latter to commercialise the technology, while enabling academia “to do experimental research rather than just working on technology or techniques that may not actually work in a real-life network environment”; it would also allow the UK to deploy 5G with more confidence and greater efficiency, “because the experience and know-how is in the country”.
Revell highlights the mobile industry’s transition from very country-specific systems to those that work on a global level, and says the UK must make sure that it keeps to the standardised version of 5G, “because if we design anything unique or proprietary, it won’t have the economies of scale and we won’t be competitive as a country”.
Both Tafazolli and Revell say the 5GIC collaborates with many countries, and while there is duplication across the world, “because everyone’s competing to come up with the best 5G solutions”, the 5GIC has a strategy advisory board consisting of its partners, and these will often come to it with details regarding what they can currently achieve in a particular area and ask it to see if it can do better. The centre takes an end-to-end and vendor-neutral perspective.
It is worth noting that the aspects of 5G that have other verticals most excited aren’t going to arrive overnight, given that 3GPP Release 16 won’t freeze until the end of 2019 – and typically, as Revell points out, it takes about three to four years after the standards freeze for there to be a ramp up in deployment. “If that all holds true then the ramp up of 5G for these types of ultra-reliable, mass connectivity will probably be around 2024, 2025.”
Revell notes that this creates issues for many start-ups looking to leverage 5G’s potential, as this long timeframe isn’t a good fit for the quick time to market they require, meaning they will need strong investors who are willing to stay in it for the long haul. He also points out that like any high-technology sector, the sheer cost of test equipment alone commands six-figure sums, creating high barriers to entry. “If you want to play in this game with 5G, it’s a significant investment before you even get to investing in people. So we have start-ups coming on campus and using our facilities because they just can’t afford some of the equipment.”
A related point is the two-speed nature of the innovation centre; once 5G is up and running, and if the industry succeeds in abstracting the complexity of the underlying infrastructure and technology, Revell expects that there will be a very fast cycle of innovation in a similar way to the current smartphone industry – where an app can be developed and deployed in a few weeks. This will require AI and automation, “because we will be creating an order of magnitude more of complexity, which cannot be managed by the conventional means we have today”.
A sense of theatre
In the same week as my visit to the 5GIC, The City of London Corporation and King’s College London organised a 5G demonstration that gave the low-latency aspect of the technology a very human twist. Musicians from the Guildhall School of Music and Drama performed in tandem with pianist Professor Mischa Dohler and the Peter Wiegold Ensemble, who played simultaneously from Berlin’s Brandenburg Gate Museum.
Latency of 20 milliseconds was achieved during the performance, using non-standard 5G equipment from Ericsson operating at 3.5GHz, which relayed the traffic onto a virtualised radio access network (VRAN) and King’s College London’s 5G core network. That then sent the signals to Berlin over the internet using a virtual private network (VPN).
Part of the thinking behind the demonstration, according to Peter Marshall, portfolio marketing manager at Ericsson, is “to show the art of the possible today; if you keep waiting then the adoption and interest from industries and consumers would happen too late”.
It turned out that the only hiccup – fairly infrequent clicks – was due to a microphone being left on on top of a piano, rather than any fault of the 5G technology.
This aside, the experience was impressive, with the musicians in different locations being able to perform together perfectly well – with a vocal duet perhaps being the most telling example. In some ways, this feat was the most impressive 5G demonstration I’ve ever seen, given its societal implications, if it can be delivered at scale. Technology has made the world feel smaller, but distance still makes itself known through the latency that anyone attempting to video-chat with their loved ones overseas will experience, and I wonder if this could help to cut the need for long-haul business travel.
Ericsson’s Marshall says 5G’s speed means that the remaining bottlenecks are in the devices and the audio compression, so one way to address the latter is to use the high bandwidth of 5G to allow the transfer of raw data without the use of compression.
According to the 5GIC’s Revell, a similar issue occurs when using 5G to stream video footage. “One of the things that we found out when we did ultra-high definition (UHD) [video], we have 8K as well, is that the video codecs aren’t good enough. We can have a communication link that has much better latency, but if the codecs can’t process it fast enough [we have a problem], which is new because it’s always been the comms [before].”
King’s College London’s Dr Maria Lema Rosas was on hand to discuss the demonstration and the university’s 5G research efforts. She says it has been trying to bring disruption via 5G to the performing arts, along with the medical and transport industries, and early in 5G’s inception spent a lot of time gathering industries’ requirements. The performance is not a one-off – King’s runs co-creation workshops with “actors, producers, directors, so they understand the limits of technology and new ways of making art”.
It has been contributing to 3GPP through its collaboration with Ericsson and British Telecom, in particular around the aggregation of 5G radio and other radio technologies, such as Wi-Fi. The university also contributes to IEEE. “King’s is leading with IEEE standards for haptic communications, which for 5G is kind of essential,” Rosas adds.
Tying it together
Professor Dimitra Simeonidou heads up the University of Bristol’s Smart Internet Lab. She says much of its focus is on the use of 5G in the urban environment. This was recently showed to great effect during its Layered Realities public demonstration in Millennium Square in March. “The temperature never went above -2°C and still we had 3,500 people outside in the square just enjoying 5G-empowered VR, AR; during the night the visitors experienced some quite spectacular visualisations of the network. It was truly great fun and we’ve been asked to repeat some of this at other events,” Simeonidou says.
Two of the use-cases that Simeonidou and her team are exploring are aimed at public safety. One was demonstrated during the Layered Realities event, using a squad of cyclists wearing helmets with 360-degree cameras (shown right). The feeds from these cameras were then analysed using mobile edge computing in combination with a machine learning algorithm to detect abnormalities, “so we were not putting the people in the square or city under surveillance, but it would generate an alert in response to an abnormal event, such as a gunshot, someone jumping into the water or violence”.
The other public safety use-case that Simeonidou brings up is trialling the use of network slicing at events such as the Bristol Harbour Festival and the Bath Christmas Market, to provide the emergency services with the vast majority of a mobile network’s bandwidth during critical events, while leaving just enough for the public to text their loved ones.
Simeonidou says the University of Bristol has also done some experimental work to compare millimetre wave and fibre backhaul methods. “Fibre definitely gives a huge advantage both in terms of supporting capacity and delivering the kind of latency that 5G needs.” She believes the UK government will have to consider the deployment of 5G and fibre together, “because they are hugely dependent on each other”.
Turning to standardisation, Simeonidou explains that while the university doesn’t directly contribute to 3GPP, it does so via its partners. However, it is working with ETSI on 5G network orchestration. As I mentioned earlier, the project funded by the DCMS was to create a 5G network that runs over all three universities’ test beds. Given that each uses equipment from different vendors, each with their own take on network function virtualisation, this was something of a challenge.
To allow this to happen, Simeonidou’s team came up with the idea for the 5G UK Exchange, which was then developed by all three universities. It allows network resources to be shared across all three test beds, and she says this is a world-first, given that it provides “interoperability across different 5G domains, and these 5G domains could have different vendors, different software stacks and so on”.
ETSI has invited the researchers to standardise the solution, and her team is currently working with mobile network operators (MNOs) such as Telefónica to do so. She adds that this capability is attracting a lot of attention from them because it is key if they want to use equipment from different vendors in different parts of a 5G network, and could allow services to be orchestrated across different MNOs’ networks.
On a much less esoteric note, as part of a project spearheaded by the West of England Combined Authority, Simeonidou and her team will be helping to roll out a test bed across major tourist attractions in Bristol and Bath, such as the Millennium Square and Roman Baths, to provide visitors with augmented reality experiences; for example, allowing them to be given a tour of the baths by a virtual Roman soldier.
Although World Cup glory continues to elude us and our politicians are burning down the Brexit clock with seemingly gleeful abandon, it is perhaps encouraging to see that a great deal of progress is being made here in the UK when it comes to 5G. The big question mark appears to be around whether we can translate this into commercial success and the efficiencies that 5G promises once deployment gets fully under way. Only time will tell…
