Mobile edge computing (MEC) is one of a number of emerging technologies that we began hearing a lot more about in 2015. MEC is regarded as one of the key building blocks for 5G, as well as being a key enabler for Internet of Things (IoT) and mission-critical vertical solutions.
So, what is it? According to ETSI’s MEC Industry Specification Group (ISG), MEC provides, ‘an IT service environment and cloud-computing capabilities at the edge of the mobile network, within the radio access network (RAN) and in close proximity to mobile subscribers’.
Dimitris Mavrakis, principal analyst, Intelligent Networks, Ovum, says: ‘The whole premise of MEC is that you need more processing capability and intelligence at the edge of the network – i.e. at the cellular base station. There is quite a bit of overlap between 5G and MEC, but they are not mutually inclusive as you can have 5G without MEC and MEC without 5G.’
The ETSI MEC ISG initiative was set up in 2014 to establish a standardised way to formalise processing capability at the edge of the network, and its white paper: MEC – a key technology towards 5G (Sept 2015) states that the aim of MEC is to: ‘reduce latency, ensure highly efficient network operations and service delivery, and offer an improved user experience’.
These performance improvements are required as mobile networks are increasingly under pressure both in terms of traffic and costs, so the industry is looking for ways to ‘maintain quality of experience, to generate revenue, and optimise network operations and resource utilisation’.
The white paper goes on to argue that this pressure is increasing thanks to the growth of IoT, which is further congesting networks, and ‘operators are looking to improve local network performance analysis to ease security and backhaul impacts’.
It continues: ‘Enterprises want the ability to enable and engage with their customers with more efficient, secure and low latency connections. Application and content providers are challenged with the latency of the network when connecting to the cloud.’
This is because while cloud platforms may have the necessary processing and storage power to handle large amounts of data, they can be hampered by the network’s ability to get that data to the data centre fast enough, especially if there are delays or connectivity issues.
These issues can be addressed if the processing is done closer to the end user – at the edge of the network in other words. So, MEC can be seen as a cloud server running at the edge of the mobile network, able to perform specific tasks that cannot be done by traditional network infrastructures or established Cloud computing platforms.
MEC is based on a virtualised platform, with an approach complementary to network functions virtualisation (NFV), as the infrastructure that hosts MEC and NFV is quite similar. They could even be hosted on the same platform.
However, Nurit Sprecher of Nokia Networks and chair of the ETSI MEC ISG, says: ‘It is very clear that MEC can complement NFV and SDN (software-defined networking) to ensure highly efficient network operations, automation, new business opportunities and a better quality of experience for the end user. But the business objective for MEC is totally different from NVF. They complement each other, but can exist independently.’
Sprecher stresses that the MEC ISG is not seeking to reinvent the wheel and it will reuse already established protocols where possible, such as NVF management and orchestration standards, for example. ‘NFV may be dedicated to MEC or shared with other functions, but MEC can thrive in a non-NFV environment,’ she asserts.
So, what does edge computing provide that Cloud computing and traditional network architectures do not? The MEC ISG’s MEC – Introductory Technical White Paper (Sept 2014) states that the typical MEC environment is characterised by:
On-premises: the edge is local, meaning that it can run isolated from the rest of the network, while having access to local resources; this is particularly important for M2M scenarios dealing with security or safety systems that need high levels of resilience.
Proximity: being close to the source of information, edge computing is particularly useful to capture key information for analytics and big data; it may also have direct access to devices, which can be leveraged for business specific apps.
Low latency: As edge services run close to end devices it considerably reduces latency; faster reactions improve the user experience and minimise congestion in other parts of the network.
Location awareness: when a network edge is part of a wireless network, be it Wi-Fi or cellular, a local service can leverage low-level signalling information to determine the location of each connected device: this enables a large number of business use cases.
Network context information: real-time network data (eg. radio conditions, network stats, etc) can be used by applications and services to offer context-related services that can differentiate the mobile broadband experience and be monetised.
The business driver for developing and investing in MEC technology is that all of these characteristics can be translated into value-creating opportunities for mobile operators, application developers, service and content providers, over the top (OTT) players and network equipment vendors. It provides a way for each of these parties to play complementary roles and to monetise, or better monetise, the mobile broadband experience.
It will do this, as the white paper suggests, by opening up ‘services to consumers and enterprise customers, as well as to adjacent industries that can now deliver their mission-critical applications over the mobile network. It enables a new value chain, fresh business opportunities and a myriad of new use cases across multiple sectors.’
Developing a standard
However, MEC will get very messy unless a standardised, open environment is developed, hence the decision to set up the ETSI MEC ISG, which celebrated its first anniversary in December 2015.
Explaining the MEC ISG’s approach, Sprecher says: ‘We started work by looking at the use cases and developing requirements out of them. The goal is to look at the evolution path and hopefully satisfy all the requirements.’
The first goals were to develop specifications for: Terminology; Service Scenarios, Technical Requirements (including use cases and their benefits); a Proof of Concept Framework leading to a Framework and Reference Architecture.
Sprecher reports that the MEC ISG has now published two specifications on Services Scenarios and a Proof of Concept Framework. Three proofs of concept have been developed to demonstrate scenarios covering:
• RAN-aware video user experience optimisation
• Edge video orchestration and video clip replay
• Radio-aware video optimisation in a fully virtualised network.
Sprecher says: ‘At the last meeting we started working on the APIs and the framework and reference architecture, where the function of all the elements has been set out. We need the APIs to show we can deploy innovative applications on top of the architecture, which will allow providers to differentiate themselves and generate revenues. And we are looking at the interfaces, too.
‘The aim is to make the APIs application agnostic so apps can sit on top of any vendor’s equipment. To do that, we need standards, definitions and agreed open APIs. We need to expedite this and ensure there is an easy entry point for developers – and this is happening,’ says Sprecher.
‘We have also set up a market acceleration group to create the right market requirements for a multi-vendor ecosystem and we are also working on additional documents covering metrics best practice guidelines and business models, but we have only just started on this.’
Sprecher adds: ‘MEC is completely recognised as a key technology for 5G. It is true that if you look at some of the use cases set out for 5G, a lot of them can be delivered already using 4G or LTE-A. But for mission-critical vertical solutions requiring very low latency and high throughputs, and for some IoT applications, we will need 5G. So, we are doing something valuable for the industry.’
There’s a number of technical challenges that need to be overcome to enable mobile edge computing, including integrating the MEC platform with the existing core network elements and user equipment.
A number of security challenges associated with the fact that IT applications are being introduced to the telecom world also have to be addressed, along with the fact that the MEC platform has to be resilient, with failsafe mechanisms in place to ensure faults do not affect the rest of the network.
The business case
But assuming all the technical issues are solved and the standards are written, Mavrakis says there is still the biggest challenge to overcome – making the business case stack up.
‘How do you justify adding new hardware to the base station when, at the moment anyway, there is no new revenue attached?’ he asks. ‘People are discussing things like LBS, augmented and virtual reality use cases, but for the moment it is unclear how this will be monetised.
‘Many operators are therefore sceptical about deploying MEC,’ says Mavrakis, ‘but there is no question that bringing distributed intelligence closer to the end user is a major trend. The other argument for deploying MEC without a tangible RoI is that it will provide cost savings to the operator. In which case, it will only be deployed on a case-by-case basis and not on a large scale, just a limited footprint.’
However, he notes it is very early days for MEC still, so it is not clear how it will be deployed. The MEC – a key technology towards 5G white paper - notes that exactly where the MEC servers are best located will vary according to requirements.
It lists a number of possible locations including: the LTE macro base station (eNodeB site), the 3G radio network controller site; at a multi-radio access technology (RAT) cell aggregation site; or at an aggregation point, which may be at the edge of the core network.
The multi-RAT cell aggregation site could be located inside an enterprise or other type of venue. In addition, ‘low-end’ MEC deployments could include a fixed network CPE server, router or controller, access point or home-based node.
Mavrakis says: ‘It is my impression that MEC may not be deployed at the base station, but at a step closer to the core network at the application point where hundreds of base stations connect before joining the core.
‘That would be less costly and much easier to deploy, as the operator can then handle hundreds of base stations at a centralised point. So, perhaps this aggregation point will be the initial deployment location of MEC and then maybe it will spread further to the edge.
‘If we extend current behaviour, particularly for small cells, and several operators want to do this, then placement and power issues and difficulties negotiating sites with councils and building owners are major barriers still.
‘The same applies to MEC if you add new capabilities at the base station. These remain practical problems for the next few years, so MEC may suffer from exactly the same issues as small cells,’ suggests Mavrakis.
Mansoor Hanif, director of RAN Development & Programmes at the UK mobile operator EE, agrees that MEC is not going to happen overnight. ‘No CEO is going to give me loads of money to do MEC unless there is clear revenue attached. But there’s a huge amount of use cases that might provide that revenue. So why aren’t operators doing it?
‘There are two types of use cases: the ones you cannot do with existing technology; and the others you can do now with Wi-Fi or existing cellular networks, but perhaps requiring some bespoke technology. What we need to do is industrialise that process and add revenue to it. In some cases it is about operators moving into new areas.’
Hanif observes that operators are naturally interested in the cost saving and synergy aspects of MEC, but adds that a network can be optimised in many different ways already, so what they are really interested in is the opportunity to generate new revenues. ‘So, if MEC takes us into new business areas that is good.
‘We can do a lot already, but latency is a challenge,’ he continues. ‘So, if you cannot get access to enough computing power and capacity in a centralised network position, you take that power and put it at the edge of network, where you get the same experience as you would with 5G by lowering the latency.’
Citing EE’s services at London’s Wembley Stadium, he says that 1,200 staff with handheld devices can be connected to the cameras, getting instant connections. ‘A normal system takes 10-20 seconds, but ours got it down to less than 0.5 seconds. That is what MEC allows us to do. There are no lack of use cases, there is just a lack of imagination.’
Hanif agrees that operators are unlikely to roll out new hardware to every base station. ‘But we have a system architecture whereby we can install MEC at aggregation points. Our belief is that we are at the edge of a massive upgrade of indoor installations with new hardware in shops, casinos, offices and venues.’
What this means is a rollout of small cells to boost indoor coverage and capacity, but Hanif argues it would be a missed opportunity to just install ‘stupid’ hardware. ‘Let’s add intelligence,’ he says. ‘What business opportunities can we add to the hardware, so that the installation and acquisition costs are low and we can build a whole wedge of services on top.’
This is certainly the MEC vision, and in fact the Small Cell Forum (SCF) has suggested that small cells may provide the strongest case for justifying investment in MEC, although there are issues associated with their computing capability and storage capacity compared with larger RAN nodes – security issues and backhaul limitations also need to be considered.
In a press release from 15 October 2015, the SCF stated that it is ‘ideally placed to contribute (to MEC) because many areas of its work in defining small cell architectures prefigure the issues which will face 5G in general, and MEC in particular – for instance in its work on services APIs and virtualisation’.
Sprecher reports that there is no lack of enthusiasm within the industry when it comes to MEC (particularly among video content providers). ‘There’s a lot of Plcs demonstrating the value of the technology, but we need to wait for the APIs to enable commercialisation. We hope to see deployment by the end of 2016.
‘Operators can then reposition themselves in the value chain and open up the network to third parties and internet providers. We will see new apps for multiple sectors and new revenue streams being generated from these new applications. MEC will open up new market opportunities,’ she concludes.