This was quite a wide-ranging event, trying to cover all aspects of wireline and wireless connectivity with multiple specialist streams including one for mobile backhaul. There was relatively little Small Cell activity overall. I chaired part of the Mobile stream and participated in two panel sessions, which included several very good presentations. Retiring European Commission VP Neelie Kroes gave a well attended speech reflecting on some of her difficulties as EU Digital Commissioner, contrasting strongly with FCC Tom Wheeler's positive speech at CTIA last month.
I saw this as much more of a wireline event, and as such there was plenty of discussion about just how fast we need our residential broadband to go. Does this really justify fibre to every home and apartment? Doug Williams of UK network BT thinks about 35Mbps should be enough to meet the demands of 95% of users in 2018. Vendor Alcatel-Lucent would rather we all had 75Mbps.
That's important because it depends on what type of technology is used to wire up our homes. Those last few hundred metres make all the difference, and it continues to amaze me just how fast those legacy copper wires can be made to go.
Figures today show that the average US household consumes 44GBytes/month of which 64% is real-time entertainment (i.e. video, gaming) whereas Europeans consume 17GB and 43%.
NFV (Network Function Virtualisation) seemed to be all the rage, with lots of demos and showcases. I suspect many are still trying to get their heads around what this actually involves.
For the (fixed) wireline services, NFV seems to be focussed around the core network and also the CPE (Customer Premises Equipment). I didn't see plans to make the DSLAMs virtual yet, which are the nearest thing wireline have to mobile basestations.
In the mobile network, virtualisation has been a great story for the core network. The term also appears to be associated with Cloud RAN for outdoor macrocell sites which we expand on later below.
4.5G – whatever that means
While the industry has been trying to figure out what 5G might mean (read our article on determining the requirements first), vendors and operators have quietly been introducing the plenty of new terminology. Perhaps they've been looking at a brochure for Coke flavours. Pick from:
The trouble is that these definitions don't align. Vodafone have been using 4.5G for Carrier Aggregation; EE mean it to be LTE+Wi-Fi (which they don't really like) and preferring to use the term 4G+ when using Carrier Aggregation.
Contrasting views on the viability of public Wi-Fi
Mansoor Hanif, Director of RAN and Programs at EE, believes that over the next 2 years there will be a seismic shift away from using Wi-Fi in public spaces. He thinks that while Wi-Fi is good in private spaces, eg at home, it's not easy to deliver a reliable quality experience using Wi-Fi in public areas.
Sami Susiaho from UK Wi-Fi operator The Cloud confidently refuted that position, describing how they had designed and optimised a Wi-Fi installation at a 30,000 seater sports stadium capable of handling 50% of fans simultaneously. They believe they can be in a position to deliver an average of 17Mbps on request 90% of the time when fans are seated. (I'd note this solution doesn't support handover as fans move around and that the VoIP services are blocked).
Edgar Figuero, chairman of the Wi-Fi Alliance, promoted their positive story of the success of Wi-Fi generally. He was also pushing Hotspot 2.0 and Passpoint, but I didn't hear many statistics about how much it's actually being used yet. The vast majority of Wi-Fi is still privately accessed and requires clunky manual login with passwords.
Small Cell Vendors on show
Although this was far from a small cell event, we did see several small cells on display as part of the wider product portfolio, tucked away in a corner of stands.
I specifically saw:
- Sagemcom, who have been successful with their 3G femtoplug (used by Vodafone as their SureSignal box) and also had on display a 4G model. They tell me that they would stick to their residential/consumer market and don't plan any enterprise products. Although they could incorporate a femtocell into their other residential products (eg set top box, broadband modem), they expect these to be standalone until the market evolves further. They don't see a requirement for combined 3G/LTE residential femto yet either. Their 4G product is still at the early development stage and the form factor could change.
- Askey, the Taiwanese ODM, had on show a 3G/LTE multimode product they have under development. They also had separate products for both 3G and 4G on display, manufacturing these for several well-known brands.
- ZyXel, another Taiwanese consumer giant, was displaying three residential Femtocells. Two were 3G, one with the option of using Wi-Fi to connect to the residential broadband modem so it could be positioned elsewhere in the home. A new one was LTE only. They have a partnership with Alcatel-Lucent and their products are certified to interwork with their gateway.
It seems that the underlying chipset for residential Femtocells today is a straight choice between Broadcom and Qualcomm.
Cloud RAN for Macrocells
Anna Pizzinat has been researching Cloud RAN for Orange. She presented some of the benefits for those trying to expand macrocell sites. In some cases, there just isn't the physical space to put more racks of equipment which then have to be powered and maintained locally. Instead, remote radio heads would be put next the antenna on the towers and connected directly by dedicated dark fibre (using CPRI) to central datacentres with racks of baseband processing.
I can't help but wonder how mobile operators, having recently spent billions refreshing all their macrocell sites, could justify replacing that equipment with centrally located racks and requiring dark fibre to every site. Arguments in favour include lower total power consumption and simplification at the cellsite. Architectures include using remote radio heads on the streets rather than true small cells, requiring very high backhaul capabilities using fibre or wireless fronthaul. A single fibre can support up to 15 CPRI links (i.e. 15 radio sectors) before needing to use DWDM (different colour/wavelengths).
It seems to me that the balance between gaining capacity through spectrum reuse (i.e. smaller cells) is radically greater than through spectrum efficiency (e.g. CoMP and other techniques). So the approach of deploying small cells with simple backhaul should be a more cost effective solution than trying to extend and expand the existing macrocell environment.
But that view wasn't shared by everyone. Michael Ritter from ADVA recognises that Small Cells are inevitable, but includes Cloud RAN/Remote Radio Heads in that scope. He believes phase timing is critical, and said that Verizon had chosen to deploy Cloud RAN rather than independent Small Cells for that reason.
In the longer term, he believes the challenges are the really big ones:
- Backhaul capacity up to 20Gbps for a full macro site
- Latency 100us between RRH and BBU (remote radio head and central baseband processing unit)
- Symmetry in order of few tens of nanoseconds
- Jitter 2ppb
- Loss 10E-12 max bit error rate
This story aligned with that from Transmode, where Sten Nordell expects fibre to penetrate deep throughout the RAN. He envisages that every radio node will be fully phase aligned using SyncE and 1588. It will use mostly passive optical infrastructure using OCUs, with (new to me) a so-called "muxponder" providing some active optical switching where required.
Kuo Huey-Lung of Chunghwa Telecom detailed a long term vision of how they will evolve their backhaul in Taiwan, favouring packet transmission using Optical networks. He likened the transition from managing older SDH technology to "making the elephant dance". He told me they typically assign 200Mbps backhaul for busy macrocell sites today, expecting this to grow to 300Mbps next year.
Perhaps it was the wireline-biased nature of the event which led much of the debate to be around using fibre. Comcast would be happy to use coax for the last few metres, and expects that US mobile operators will continue to outsource backhaul connectivity to them. I also spoke with one European operator that was looking at using copper for the last mile to urban small cells – its relatively quick and easy to deploy – but may not provide all of the phase sync timing required.
Delegates at this event could come away with an unbalanced view of how mobile networks will add capacity through densification. The strong case from those promoting fibre and Cloud RAN everywhere wasn't counter-balanced enough by those advocating wireless backhaul and traditional small cells. I felt I was one of the few positioning this alternative approach and educating what alternative wireless backhaul products can do today. DAS vendors were also largely absent. Enterprise/in-building wireless wasn't really addressed.
It is clear from a few conversations with delegates that small cells are actively being trialled by operators worldwide. Solutions for outdoor urban densification looks like being a battle between street mounted small cells with up to 100-200Mbps backhaul and Cloud RAN macrocells with a few remote radio heads connected by up to 20Gbps dark fibre.
Next year's show will be in London.