London was again home to the 15th annual Transport Networks for Mobile Operators (TNMO) conference. As with quite a few events these days, there was a strong theme of small cells which were mentioned in the majority of talks. Also in common with other conferences, I found a wide range of quality of content – from the outstanding to the truly banal. New takeaways for me related to small cells included the work being done by the MEF to determine the characteristics of Ethernet required for LTE small cell backhaul, various offers of Small Cell Hosting and the (mostly private) comments from operators about the reality of Metrocell rollouts.
We've also recorded a short (3 minute) vox-pop style video from the chairmen and several speakers during the event to give you a sample of the atmosphere.
Rather than give a blow by blow account of each presentation, I'll focus on three key themes:
"Small Cell as a Service" vs "Small Cell Hosting"
The UK seems ripe with a number of companies offering their services to mobile operators. Some already provide wholesale backhaul, some have extensive fibre/transmission networks in the ground, some have their own wireless spectrum. Strangely, I haven't seen similar activity in other countries, although one delegate did point out a couple of US companies.
MLL Telecom operate a number of (mostly wireless) backhaul links for macrocells, owning some microwave spectrum and having their own field force. They offer both expertise to survey, deploy and manage small cell networks – capitalising on their existing NOC (Network Operations Centre) and radio network expertise. CTO Peter Jennings believes that wireless links can be cost competitive over fibre across distances of anything up to 15km.
Virgin Media provide wholesale wireline backhaul using their extensive cable and fibre broadband. They discriminate between the terms "Small Cell as a Service" and "Small Cell Hosting", offering to install, backhaul and maintain any boxes that an operator wants. They already have good relationships with civic authorities and know how to dig up roads. Kevin Baughan, Director of Metro Wireless, holds a view that fibre runs of more than about 15 metres from the existing point of presence were not commercially viable, preferring a wireless link for the last few hundred metres. They've been working on the Small Cell Hosting concept for a couple of years, have done trials, and also rolled out Service Provider Wi-Fi service which establishes a footprint on which Metrocell (i.e. cellular) functionality can be added.
COLT, a European business broadband provider, delivered the most technical presentation, covering details about timing and synchronisation which demonstrates they have put a lot of thought into the subject. Whilst their trials are not as public or visible as others, it is clear they are highly active in this space.
BT were perhaps notable by their absence. Their recently acquired 2.6GHz TDD spectrum was mentioned by other speakers. When combined with their nationwide presence, huge wireline network and existing wholesale backhaul services, they could offer a seriously competitive alternative to others.
Ethernet requirements for LTE small cell backhaul
Both one Ericsson and two MEF speakers touched on very recent work being done by the MEF (Metro Ethernet Forum). HetNets and LTE-Advanced features in particular place some very demanding requirements on the underlying backhaul, which they have categorised into three or four classifications.
Firstly, 3G has nothing to worry about on that score – each macrocell and small cell only needs to have frequency alignment.
The initial LTE deployments can also be adequately handled using existing technology, albeit higher speeds. Several delegates talked about typical standalone LTE macrocells needing 1Gbps fibre backhaul. Many will be upgraded to 10Gbps, not because they'll use the full capability but partly because that's the next step up in transmission size and partly because nearby small cells would share the capacity.
Some of the most advanced LTE features require very close alignment between multiple cellsites. CoMP (Complementary Multi-Path) allows the same smartphone to be talking with both a macrocell and small cell simultaneously, squeezing the most out of the radio spectrum for both uplink and downlink. eICIC (enhanced Inter-Cell Interference Co-ordination) synchronises the quiet times of macrocell and small cells in turn to reduce interference and increase system performance. There are various levels of how closely the cells work with each other and the backhaul delay and phase synchronisation needed.
|Co-ordination Level||Use case||Phase Sync
|Moderate||picocell in a macro network||+/- 5us||None|
|Tight||picocell in a macro network||+/- 1.5us||1-10ms|
|Very tight||joint scheduling between multiple cells to same device||+/- 1.5us||<0.5ms|
This leads to performance demands of latency which can't be met using traditional Ethernet technology. For very tight co-ordination, either dark fibre or specialist wireless running CPRI would be needed – to my mind, effectively operating more like a remote radio head than a standalone small cell.
The relevant specification to be written is MEF 22.2 which will cover Small Cells with tight co-ordination backhaul and is backwards compatible with the MEF 22.1 spec available for today's 3G and LTE networks.
So operators can start with 3G backhaul today, then add phase sync in a year or two then later. The major benefits of features such as eICIC (to reduce interference between macro/small cells) aren't worthwhile unless you have several small cells per macro, and sharing the same frequency.
In their presentation, Ericsson also took the view that some of these advanced LTE features would require fibre connections to all the cells involved.
I'd argue that the most demanding aspects of some of these LTE-Advanced features change the backhaul requirements so much that they can't really be called Small Cells anymore – it's more akin to the Cloud RAN or Remote Radio Head architectures.
Will operators deploy 3G or LTE metrocells first?
Given that this was a backhaul network conference, delegates are looking at public access metrocells rather than residential or enterprise sectors. Publicly, we heard several operators say they would focus on LTE small cells first, introducing these after their initial LTE macrocell coverage deployment in order to provide capacity. TIM Italy clearly stated this intention, although qualifying it for countries such as Argentina where they don't yet have 4G spectrum.
Privately, I heard other views, with one operator convinced that the large installed base of 3G users will continue to generate the bulk of their traffic demand – not everyone wants to pay $100/month for 4G service if 3G HSPA+ meets their needs. They could see customers were choosing to downsize, retain existing phones on SIM only contracts rather than upgrade to the latest LTE as before, but use greater amounts of data.
Another operator told me they remain concerned about the high cost of outdoor small cells, and would like to see that driven down – they don't expect to see significantly greater ARPU (monthly revenues from customers) from LTE, thinking in terms of perhaps $5/month more, and so need a compelling business case.
A couple of operator speakers (e.g. Telecom Austria) were keen to point out that they weren't see a data tsunami of epic proportions, but instead a more linear data traffic growth which they felt they could more than manage. Clearly, there isn't consensus on future traffic forecasts.
- FON, the crowdsourced Wi-Fi sharing network has recently signed up T-Mobile/Deutsche Telekom and will be expanding into another 6 networks. They weren't able to comment on how (or if) their solution will be incorporated into Wi-Fi enabled small cells.
- Gilat, a satellite backhaul equipment vendor, forecast up to 4% of cellsites worldwide might be satellite connected in future, enabled partly by rural small cells.
- JDSU launched their JMEP (JDSI Micro Ethernet Probe), a small dongle style device which allows remote monitoring and testing at Ethernet ports across the network to test/validate end-to-end performance.
- Intracom have incorporated mechanical self-alignment into their 28GHz microwave PtMP box. Aimed at street level deployment, it achieves a compact form factor and uses the same hardware for both ends of the link.
- Ceragon, the large established wireless backhaul vendor, has classified Small cell backhaul into four types (Offload, Integrated, Co-ordinated or Distributed) and mapped these to a mix of technologies based on cost/capacity tradeoff. Visit their website for a white paper on the topic.
5 minute video
We recorded a few minutes video highlights including interviews with the chairmen and several speakers.