For the rail industry, how to replace the ageing GSM-R radio networks is an ever-present challenge. Thoughts over the past ten years have progressed from 3G to 4G LTE but increasingly nowadays to the prospect of the soon to emerge 5G technology. With its huge increase in data capacity, a 5G rail network should be able to meet all rail usage requirements including operational, commercial and passenger communication needs.
A paper given to the IRSE back in December 2017 jointly by Rail Engineer writers Clive Kessell and Paul Darlington looked at the options for replacing GSM-R and concluded that 5G was likely to be the solution when GSM-R obsolescence becomes a real problem in the mid-2020s.
But just how far off is a general roll out of 5G networks? A recent seminar held in London organised and hosted by Cambridge Wireless explored the present position from three different perspectives:
- A mobile operator’s view;
- The emerging business models and trends;
- An equipment provider’s view.
From these emerged some surprising elements and it is clear that a number of conflicts of interest remain to be resolved before, and indeed if, a united view can be agreed as to how 5G should be structured.
The seminar focussed primarily on the forthcoming 5G public networks, and there was little mention of ‘mission critical’ services on which the railway and other industries such as the armed forces, the emergency services, nuclear power generation, essential utilities and others would need to be reassured. Nonetheless, the challenges within the public offering will have a bearing on how mission-critical services are structured if 5G is to be the technology of choice.
5G vision and characteristics
The advent of 5G goes hand in hand with IT developments that require reliable and high bandwidth connectivity. Foremost of these is the Internet of Things (IoT), but also the insatiable demand from the user community for ever-more applications to be available to end user devices. These culminate in a system that aims to provide i) between 10 and 100 times more connected devices, ii) 1000 times more data volumes, iii) up to 100 times end user data rates. Connectivity of 10Gbit to end devices is a possibility.
To fulfil this, the vision is for a cloud computing architecture with software-defined networking. A Centralised Radio Access Network (C-RAN) would enable all suppliers to access the cloud using clusters of LTE base stations. Uplink signals from multiple base stations optimise network configuration by selecting the best signals at a number of receivers, thus minimising the effect of interference from other operators.
Virtual networks for different user groups are capable of being created, by the concept of ‘network slicing’, to enable a degree of separation within the cloud. Small cells are envisaged, which not only means much-improved radio connectivity within buildings but would allow end user devices to connect directly with each other. This aligns with the concept of ‘mobile edge computing’ that pushes the core functionality out to cell sites. User equipment would typically connect to multiple cells.
All of this requires radio spectrum, meaning that ever-higher frequencies are needed to fulfil the need – typically between 20 and 60GHz will be allocated.
5G technology development is substantially completed (up to release 15 on the standardisation programme with release 17 seen as the effective end point) and has reached the point whereby equipment will shortly be produced commercially.
A mobile operator’s view
The BT/EE view, as put forward by Philip Bridge who has responsibility for network architecture, shares the vision for what 5G is capable of offering but has significant doubts as to how this will be realised. A new core will be needed to achieve the benefits of infrastructure decoupling, virtualisation within the cloud and the introduction of new services. 5G is being considered in three different ways; a ‘Network View’ a ‘Functional View’ and a ‘Deployment View’.
How to integrate these three perspectives is potentially very difficult. Currently, a silo mentality prevails in the supplier community, thus making development of a ‘common cloud’ something of an alien culture. There will also be a need to integrate 5G with the ‘long tail’ of legacy equipment in 2G, 3G and 4G networks and provide roaming between all of them. The need for interworking boxes may have to be considered although this is not a desirable way forward.
4G networks are already being built on a 5G basis, thus making the eventual solution easier. Even if established, the instrumentation and training required to achieve excellent network performance is not easy with a cloud-based architecture. Methods used today for network monitoring will not work for the 5G vision and thus a transition has to happen, but with no means of knowing how this can be achieved.
It is considered that very few vendors have the stature to build a network of this type and a multi-vendor solution may have to be forced by some form of legislation. Infrastructure sharing is a likely way forward with ‘Telco Grade’ networks setting the required standard. A combined service orchestration to break up the silos with a simpler but more stable architecture will need putting in place.
The goal is full interoperability between vendors A, B and C in both hardware and software, which will be difficult to mandate and manage in terms of who would be in control of exactly what. If not executed properly, matters may become worse when compared to the present service provision, particularly in terms of faulting, maintenance and associated payments.
Whilst the technical elements are challenging, the contractual and pricing structures between the different service providers are going to be equally difficult. The declared intention of catering for autonomous vehicles makes it essential that mission-critical services are encompassed within the cloud. Such services would require priority over other applications in public 5G networks, but exactly how this will be achieved and at what cost is a question still being posed. Railways take note.
Emerging business models
Having listened to the probing view from a mobile operator, a more upbeat message comes across from the ‘forward thinkers’ within the industry. Alan Carlton from InterDigital Europe posed the question as to the role of the Core and Edge elements of a 5G network, the latter being in the ascendency but recognising that the core will still be important. A service-based architecture (SBA) is emerging to satisfy both the IT and telecom requirements leading to a flexible data centre approach that somewhat blurs the core and edge debate.
Quality of service should be at the forefront of future thinking, says Mischa Dohler, professor in wireless communications at Kings College London, and at present it remains somewhat doubtful. When comparing radio spectrum to Wi-Fi, the latter, according to statistical research, is more reliable, which is surprising since Wi-Fi operates in an unlicensed part of the spectrum. The development of radio networks has mirrored, to some degree, the architecture of 2G, which existed before the internet was invented, hence the problem.
One factor that stands out is that the UK is awash with fibre that is not being used efficiently. Connecting this all together on a shared basis would be a big asset for future radio development, in that it would make networking that much more resilient.
Localised 5G services are already in being, the city of Bristol network trial being described by Dimitra Simeonidou from Bristol University. Offering both LTE 2.6GHz and Wi-Fi services, the network uses a combination of municipality and infrastructure owners as 5G neutral hosts.
Designed to be vendor agnostic, three suppliers – NFV with an Open Source MANO platform, SDN with its NetOS controller and Nokia’s Cloudband and NetAct controller – have each supplied products to create the network to produce cloud and radio platforms out to the extreme edge of the city. The system exists as a reality, but is there primarily to prove that a 5G architecture of slicing, convergence and collaborative hubs can be constructed.
The supplier’s perspective
That suppliers are working hard to develop 5G equipment is not in doubt but the end game, in terms of what the products will deliver, is an ongoing conundrum. David Astuti from Huawei gave some predictions. The customer experience should yield a 10-fold benefit with 15Gbit/user being possible. The efficiency of new application releases should reduce from typically six months to one week, although some delegates disputed this. Connectivity is expected to improve by a factor of five.
The ‘slicing’ concept will enable different slices to be used by different user groups. Typical of these will be autonomous driving (already mentioned), smart campus networks and home domestic controls.
Getting from the present to the future needs a migration strategy. A pragmatic way forward is to:
i) Continue with 4G expansion but make this 5G ready;
ii) Create standalone 5G core networks based around a network service architecture;
iii) Merge these to create a converged core 5G network.
2018 should see the first standalone core networks introduced in readiness for such events as the 2020 Tokyo Olympic Games, with converged networks not likely to appear until sometime after that, except for local applications.
It became clear that 5G, from a technical perspective, is making good progress, but the commercialisation of 5G services will need much more work and much more openness between suppliers.
The subject of cyber security will need to be a key priority, with Nokia believing that this will be a fundamental element of the core network.
5G must be able to handle traffic from legacy networks of 3G, 4G and even 2G origin. Just how this will be achieved is not clear. The Internet of Things and particularly the Industrial Internet of Things will be dominant requirements in 5G connectivity.
Spectrum allocation is akin to a power supply – if it is not there, the system will not work. 5G services will use a wide range of frequencies as it is envisaged that 5G products will connect to services rather than individual radio channels. Unlicensed spectrum will be part of this – local city networks will likely operate in unlicensed bands – bringing its own management challenges.
Mission critical services will, in time, inevitably move to 5G, so arrangements need to be made for these. Special facilities such as group call, push to talk and location dependence operation, important for both emergency services and rail operations, are part of the requirement.
Convincing the safety authorities that a common user 5G radio bearer is suitable for control and command will be a sensitive issue for the rail industry authorities. Whether or not rail can have a dedicated ‘slice’ with its own allocation of spectrum remains to be seen.
Attending a radio seminar not focussed just on rail was an eye opener, as it put the rail interests into perspective. It will be interesting to see what happens over the next two years or so and, particularly, whether the suppliers are minded to work for greater co-operation and integration.