Swimming with the tide of 93,000 delegates attending MWC this year, it was easy to become overwhelmed by the sheer scale and breadth of the event. It also felt that the industry agenda is no longer driven by mobile network operators, who are themselves being overwhelmed and/or overtaken by the much wider ecosystem that they enable. A huge array of wireless enabled products and services were on show, some of which seemed only marginally related to cellular. The lines are blurring and you can't retain a silo mentality. This report tries to focus on small cells but inevitably spills over to related topics.
Network investment today is very quickly shifting across towards 4G. Huawei forecast around $8B of their $14 Billion revenue for 2015 will come from LTE. Many of the new small cell products and features we saw last year have matured and become commercialised. 4G small cells are widely available including many 3G/4G configurable and a few with LTE carrier aggregation. Wi-Fi has also continued to evolve, especially Voice over Wi-Fi, leading to debate about how best to make both work together. Meanwhile, the last year has seen a lot of good common sense engineering and operational improvements going on in the background that are easily overlooked. The range of small cells on display at the Small Cell Forum booth (right) continues to impress.
There was a lot of optimism in this market segment, which is definitely growing. I sense that with many initial LTE network investments now well advanced, operators are finally finding time to respond to in-building opportunities – there's certainly strong demand for it from venue owners. Derek Peterson, CTO Boingo Wireless, provides a lot of neutral host in-building systems across the US. He told me they use a mix of DAS, Wi-Fi and Small Cells to address the needs of each venue – there is no single technology that fits all requirements. He's not just providing wireless connectivity to solve the needs of network operators – venue owners value features such as aggregated traffic reports showing where and when the bulk of visitors move around and where they linger.
Addressing that same need, Cisco launched MobilityIQ which provides those smart analytics across both cellular and Wi-Fi, aimed at the building owner rather than the network operator.
The big surprise was Cisco taking on the Spidercloud E-RAN system, complementing their own standalone small cell products and targeted at the medium to larger buildings. This is not just a reselling agreement, Spidercloud radio nodes have been engineered to plug into Cisco's Wi-Fi access points (single channel, remotely configurable for 3G or 4G) and interwork with Cisco's small cell gateway. Publicly approved by the Vodafone, Verizon and others, this enables their partner SI channel to go out and install.
ip.access also reported healthy activity at the show. Their Enterprise range includes a fully commercial LTE only product (I'm told now shipping to 22 customers). They are the only independent small cell vendor supporting the full set of 2G, 3G and 4G technologies. Elsewhere, Alcatel-Lucent have progressed their Qualcomm based 3G/4G multi-mode product. Ericsson announced that Vodafone Netherlands retail outlets will be the first use their new RBS6402 multimode small cell we first saw at CTIA last September. Quortus added VoLTE support to their enterprise core network software.
On the planning and deployment side, iBwave reported adding another 3,000 parts to the comprehensive product database used by their tablet planning app. Ranplan launched their iBuild Tablet Planner to address the same market. Spidercloud have embedded a Bluetooth link into their radio nodes to make them easier to maintain (can connect and diagnose even if the LAN is broken) and can also be used to find them when hidden behind ceiling or wall panels. Airvana launched a small cell management platform that copes with large scale multi-vendor deployments. XCellAir are a startup that interworks their own SON for Wi-Fi with 3rd party SON (e.g. Airhop) for small cells. They combine that with a scalable management cloud-based system that works on top of or in place of existing management systems.
Many products are now remotely configurable between 3G and 4G. Some markets, such as the US, are seeing demand for 4G only as they embrace VoLTE – Verizon is a good example. Martin Guthrie from NEC thought that Europe was still more predominantly 3G, noting that Turkey (where they've just announced a win with Spidercloud) hasn't yet issued 4G licences.
Small cell innovation was recognised by the GSMA firstly for Airvana who won two awards for OneCell – Best Mobile Technology Breakthrough and Outstanding Overall Mobile Technology. Secondly, Huawei won the GSMA best Mobile Infrastructure award with their Lampsite Enterprise product. Already deployed in more than 70 operators, it delivers 3G, 4G and Wi-Fi for larger buildings and is compatible with their macrocells. After two years commercial deployment, I'd say this was well ahead of the comparable Ericsson Radio Dot. I was surprised to see this architecture being used by small cell vendor Contela, who were showing a new SIS product on their stand – fibre to a hub then CAT5 cable to each radio node.
Some TD-LTE enterprise products are also coming through. China promises to be a very large volume market but the immediate demand has been for Japan. Not all chipset suppliers support this mode yet, but once they do it's a relatively easy feature for small cell vendors to introduce. Contela had a 64 user TD-LTE product and also have a 96 user enterprise product on their roadmap. Aritel, another South Korean vendor, has also been exporting TD-LTE Enterprise small cells to Japan.
Microsemi launched a new inbuilding timing product which uses GPS to generate a master clock source, essential for high performance LTE-Advanced features. You can register for our webinar on this topic on 31 March.
This is perhaps the most complex market segment of small cells to assess – there are many, many different technical solutions on offer. While there's little doubt that densification will mean more radio heads down at street level, there are lots of ways to achieve that. The mainstream RAN vendors would encourage use of their own equipment for compatibility, either with picocells or remote radio heads. Challengers and small cell vendors would propose a separate underlay of fully independent small cells.
Remote radio heads need 2.5Gbps or more per channel, and are really only feasible where fibre is easily/cheaply accessible. eBlink support that wirelessly, handling up to three CPRI links on a point-to-point basis and are endorsed by Orange. Cavium demonstrated a different architecture which added much of the latency intolerant processing to the dumb radio head, reducing the fronthaul capacity to about 200Mbps. For comparison, a fully standalone LTE small cell should only need 100Mbps or so.
The larger RAN vendors recognise the need for LTE urban small cells in due course and all have picocell products today which they will evolve. They just don't see the urgency and I think want to sweat the macrocell assets as much as possible first. At their pre-MWC press briefing, Huawei pre-announced an EasyMicro product for 2017 (2x5W, 2x20MHz, 6 litres, 6kg, integrated antenna) to supercede their Atomcell. They believe they already have a product set that meets today's market needs but would revise their roadmap if market demand changes.
Gordon Mansfield, EVP at AT&T Wireless, has been reassigned from Small Cell champion to lead their Access Analytics group. It's very complex figuring out exactly where and how to invest in new capacity. ATT have developed their own HARP tool for this purpose and actively use SON to squeeze the best out of the network. Macrocells are very inefficient when serving customers at the cell edge; using HetNets with small cells can reduce resources required by 19% on average. When I asked if he also took account of Wi-Fi usage, it seems that's just too much complexity to handle at this stage. Devicescape told me later they can supply "Pseudo-CDRs" to help planners visualise Wi-Fi usage on these planning tools, but I've not heard of many operators yet taking such a holistic view and fully including Wi-Fi as part of the RF planning process.
Self-Organising Networks (SON) is important to maximise network performance in an urban HetNet. Independent SON vendors such as Celwize argue that 3rd party SON is essential in a multi-vendor environment. They won Telefonica UK which because of the Beacon network sharing arrangement actually involves managing both Vodafone and Telefonica networks across half the country. I've not heard of a multi-operator SON deployment before – perhaps this might become more common as shared in-building systems evolve in future.
TTP demonstrated the concept of installing small cells on top of existing light poles/lamp posts. I hadn't realised the light sensor (to detect when to turn the light on/off) is mains powered. Simply replacing the existing sensor with a small cell solves the issues of location, power and relatively unskilled installation. A wireless backhaul link could be fixed lower down the pole and powered over Ethernet. A more expensive option is to install a dedicated pole such as Enersphere's ePole which can serve multiple purposes, from CCTV to Small Cell.
On the backhaul side, the nascent wireless small cell market hasn't been mature enough to fund an entire company. I've heard this from most of the backhaul vendors at the show, who are cautious about market predictions but keen to be able to "catch the wave" as the opportunies arise. Dragonwave thought the market for wireless urban small cell is still about a year away – operators would try to use fibre first – but when it comes, the capacity required would be around 300Mbps per node to cater for LTE carrier aggregation. Meanwhile, some early startups seem to have fallen by the wayside while most others address multiple markets.
60GHz V-band remains a good first point-to-point choice. Siklu announced a Carrier Wi-Fi backhaul deal in Baltimore. Samsung Networks launched their own product.
Point-to-multipoint systems are good at providing area coverage. CBNL have increased the bandwidth at a central point to 14Gbps by increasing the number of sectors and doubling sector capacity. CCS, who use a mesh multipoint-to-multipoint architecture, are now shipping commercially. Each of their nodes includes a GPS receiver which provides a by-product of accurate phase timing that can be distributed throughout their system, even where there is no view of the sky.
RADWIN, a big player in Non-Line-of-Sight backhaul, announced wins with multiple (unnamed) Tier 1 operators.
You could easily have thought the residential femtocell market was dead because Voice-over-Wi-Fi (VoWiFi) has been introduced into the latest iPhones and Androids. This provides a seamless voice call experience just as if you were on cellular without the need to use a separate App or address book.
Instead, we're hearing of substantial orders being shipped for femtocells in 2015. Even T-Mobile US, the "Un-carrier", are shipping 4G femtocells to their customers if they want one. I spoke to one happy customer who had switched across from AT&T to save money and found the device worked seamlessly. There's little public information about it – one website shows a possible early prototype alongside Nextivity's repeater product.
Zyxel continue to ship significant volumes of 3G and support 4G trials and see a strong future for both. They compared Wi-Fi voice with cellular by framing it from the customer viewpoint. "If I'm choosing Wi-Fi, I consider it free, so can accept it's sometimes of a lower quality. But if I'm paying for a voice service, such as to my cellular network provider, I expect good quality. I don't really care what radio technology is used, it's the end result that counts." They offer enterprise grade Wi-Fi as well as small cells and market a range of products to match their customers' specifications.
While VoWiFi may be ideal when only one or two are attached to an access point, congestion and interference can disrupt in the busiest environments. For example, I wasn't aware of anyone making VoWiFi calls at MWC; even some of my text messages didn't get through. Edgar Figueroa, CEO Wi-Fi Alliance, believes that VoWiFi can be made to work even in congested areas and assured me that "we are not yet done with Wi-Fi evolution".
The big unknown in this year's residential forecasts is the UK, where BT were expected to rollout huge volumes of 4G to their fixed broadband customers. It's unclear how BT's acquisition of EE will affect that. One supplier I spoke to remained very positive that the rollout would start this year.
TalkTalk, a UK fixed broadband network and "thick" MVNO, explained how they could use their GSM guard band spectrum at 1800MHz to launch an indoor LTE service. Their research proved that the unusual 3.3MHz bandwidth can deliver VoLTE voice and data up to 16Mbps and works with today's popular LTE phones (they tested with about 30 models). It also provides coverage in the street outside up to 60 metres away – much better than Wi-Fi. Without any other interference in that band, they reckon they could cover neighbourhoods with as little as 7% penetration. It's still early days – they haven't decided on the physical product form factor, would require formal approval from Ofcom, pricing etc. They appear to be aiming for a similar business model to that of Free France, but using nationwide spectrum that originally cost them around $250,000. A bargain.
RURAL AND REMOTE
The Small Cell Forum release 5 was issued at the show, covering rural and remote applications. There had been a perception that this was very much a small niche, but they found this wasn't the case. There are also over 17,000 small cells deployed in a wide variety of use cases. Real Wireless sized the business opportunity to be worth $163 Billion and address 650 million people.
To emphasise the scope includes transport, BMW gave a presentation showing a small cell fitted inside the car. Just as buildings are constructed from materials that are both thermally efficient and RF isolating, so are cars. By using an antenna on the roof, like we did in the early days of mobile, much better performance and coverage can be achieved.
Looking ahead, Alan Law, SCF Chairman, explained that Release 6 in June will cover Network Virtualisation. The Forum is currently reviewing their longer term release plans, but there is no shortage of topics to work on. There will also be greater regional focus, with regional champions announced throughout the year.
Powering these remote and rural sites can be a big issue and in developing countries can cost anything up to 40% of the total network operating cost. PowerOasis combine the latest Lithium-Ion batteries with solar panels and/or turbines and a clever energy management system. Add in a small cell and some of Quortus latest core network software and you'd have quite an efficient solution.
LTE-U and Wi-Fi Link Aggregation
There was huge debate about how best to use unlicenced spectrum, especially the Wi-Fi bands at 5GHz. There are two main options, clearly explained on the Qualcomm website:
- LWA (LTE Wi-Fi Access) where an LTE data session is augmented by extra traffic sent over standard Wi-Fi. This can be aggregated in the core network so could potentially travel over completely different Wi-Fi and small or macrocells. Qualcomm's expert showed me that they see the Wi-Fi path always traveling through the nearby small cell (see diagram below)
- LTE-U (LTE in Unlicenced bands). This will need some modification in the way LTE works in order for it to work alongside Wi-Fi. A Listen-Before-Talk feature would probably be needed, which may reduce some of the technical advantages of LTE.
Both approaches are work items for 3GPP Release 13, although LAA seems to be further advanced. Alcatel-Lucent announced support at the show for both, thinking that LWA could be implemented quite quickly as a software update for most smartphones.
As you can imagine, there were some very strong views on this. Speaking at a conference panel session, Simon Brown (CEO ip.access) reflected the views of many by saying that a strong regulatory and standards framework must be in place to ensure fair play for both. He thought that carving out dedicated spectrum at 3.5GHz may be a better place for LTE-U. Dan Rabinovitsj, COO Ruckus Wireless, pointed to the fast pace of Wi-Fi evolution which has been accelerated because of its unlicenced approach. He doesn't want Wi-Fi to be held back by having to consider wider interworking implications or work through 3GPP standards processes for each new feature.
At 5GHz, this scheme is mainly for in-building use and so would be implemented within Enterprise Small Cells. Ericsson and Huawei have both said they could implement LTE-U quite quickly on their products, but it would require hardware changes in the smartphones. Alcatel-Lucent believe that LWA could be implemented with just software updates and achieve much of the benefit relatively quickly.
Photo below taken on Qualcomm booth - there are other alternative architectures for Wi-Fi Link Aggregation too.
SK Telecom presented their view of 5G small cells, which involved augmenting an existing LTE data session with even higher speed wireless data channels, using large bands at anything up to 100GHz. It seems to me this could be quite achievable in a relatively short timeframe, with 802.11ad Wi-Fi (formerly Wi-Gig) a good short term candidate to achieve 1Gbps or more inside a room.
OTHER INTERESTING AND UNUSUAL STUFF
A few companies that are stretching the laws of physics:
- Kumu Networks continue to astound. They reckon they've cracked the ability to send and receive RF at the same time and frequency, effectively doubling capacity. Since we last saw them, SK Telecom and Telefonica have extensively tested and agree. One quick route to market is to use this extra capacity for small cell backhaul, using a small amount of macrocell resource to achieve high speeds to an end user that was previously at the cell edge via a small cell.
- Huawei's Surface Micro is a concept small cell of similar format to a large flat screen TV. It has many integrated antenna with beam forming and could be extended up to 128 MIMO. They've already tested this with 16x16 MIMO and achieved 1Gbps. They think this concept could be extended or adopted within 5G in the future.
- Major vendors were showing various methods of achieving 600Mbps or more. Qualcomm did this with 256QAM and Carrier Aggregation. It seems to me there is still plenty of innovation left in 4G to keep engineers busy for a good 10 years or more.
- Google announced they would be getting into the mobile service business themselves, perhaps even launching small cells in the sky using tethered balloons and solar powered drones. They hope others will copy their innovation, but aren't revealing too much as yet.
[Figure above: 584Mbps using Carrier Aggregation and 256QAM, demonstrated on the Qualcomm booth.]
Apologies for omitting many details and announcements from the show due to pressure of space. I'll be following up on many of the topics covered in this report and adding more detail in upcoming weeks. Let me know what I should be including or expanding on - just comment below (can be anonymously).