Perhaps the GSMA felt it was time to remind the industry what traffic congestion might feel like. Getting to and from the new venue for Mobile World Congress seemed to be a choice between long delays (the taxi queue), going offpeak (7am was quiet), offloading (onto buses) or compression (into tightly packed trains). The frustration of the 80,000 people was plain to see, despite valiant attempts by the organisers to handle traffic flow (a attempt at policy management) and the presence of security patrols. Even the VIP priority entrance round the back looked a bit congested to me at times.
Inside the new building, it was very spacious and felt more like a large airport terminal with the associated long walks between stands. Wi-Fi worked quite well this year, (arguably better than the cellular network which dropped calls), reinforcing its potential (for indoor use anyway) and made more attractive to visitors by the high cost of mobile data roaming (emphasised by Verizon's CTO still looking for their first 4G roaming arrangement).
Indoor vs outdoor
That outdoor vs indoor experience was also relevant to the small cell vendors and products on display. There's perhaps a clearer demarcation between those focussing on indoor (especially residential and enterprise) vs the higher power/more complex outdoor solutions. It' also noticeable that most small cell vendors are very much either LTE only or 3G today, with a few introducing the "nivarana" tri-mode 3G/LTE/Wi-Fi boxes that would be ideal in the longer term and one or two switchable 3G or LTE boxes on show.
Outdoor small cells of a sensible size are now appearing on all the major RAN vendor stands, including Ericsson, Huawei, NSN rather than just Alcatel-Lucent's metrocells. These are getting down to the inconspicuous "burglar alarm box" size but are not all the same from an architecture or design perspective.
[Small Cells need to work outdoors too - delegates enjoying a brief bit of fresh air and sunshine during the show]
While I was expecting to see/hear a lot about outdoor metrocells, it was surprising to hear/see so much about indoor enterprise solutions. Martin Guthrie of NEC reported increased business activity for enterprise systems. Traditional DAS solutions are not just expensive, but take many months to deploy and operators can only meet 20% of annual demand. There is still some disbelief that a Spidercloud solution can really be installed onsite in a matter of days, but the reality is that this allows the same budget to stretch further and satisfy a larger proportion of customers. It seems operators are looking to the enterprise segment to bolster revenues – installing small cells helps reduce propensity to churn. Spidercloud themselves stress the range of upsell opportunities that can build on an initial coverage solution. They also see the same segmentation found in the Wi-Fi market (e.g. low cost residential vs industrial strength enterprise) reflected in the range of small cell products on offer today.
Cisco revealed perhaps the biggest small cell surprise at the show, launching a 3G femtocell plug-in module that fits into existing enterprise/public Wi-Fi access points. (Perhaps they should ask to fit them into the Wi-Fi boxes at the conference next year to improve the cellular service.) This is very much part of a wider Wi-Fi/cellular integration strategy including the latest Next Generation Wi-Fi Hotspots that takes the best of both technologies to produce a more comprehensive and well thought out solution – the arithmetic of 2+2=5 applies here.
Small Cell Forum focuses on teaching operators how to deploy small cells
While 3GPP seems content with pushing out a release every year or so, the Small Cell Forum announced plans to generate three this year alone from a standing start – one each for residential, enterprise and metrocells. These releases are touted as a "How To" guide for operators who haven't yet dipped a toe in the water, explaining both the business and commercial aspects for each category based on the wealth of experience from the 46 operators (and even more vendors) who have launched commercial service to date.
A quick glance shows this first release to be a 40 page overview document that summarises and directs readers to the various existing publications to date. I didn't feel this was quite the "Residential Femtocell Rollout for Dummies" tome that the press release might have indicated, but more of a "lonely planet guidebook" to the topic which highlights those sometimes overlooked but worthwhile documents and navigates the reader through to the chosen destination. This is an ongoing program and I'm sure will develop further, driven by the new operator oriented leadership of the Forum.
View the Small Cell Forum document library here
While the Forum remain bullish about the growing number of small cell operators listed in their latest market update, ip.access point out that they alone have increased their customer base by 40% during 2012, from 100 to 140. Although not all customers are full mobile operators, it indicates that the Forum's 46 operator total may actually be an underestimate. They are the only vendor offering 2G and 3G (soon also 4G) using a single gateway and management system.
The variety and range of products from Small Cell Forum members continues to grow., for example the Metrocell top left, satellite transceiver top right.
The need for speed
Operators seem to be keen to achieve ever higher peak data rates, with Carrier Aggregation (bonding traffic across multiple LTE frequency bands) being an extremely popular LTE Advanced feature that will be "the next big thing". This bonds separate LTE frequency bands/carriers transmitted by the same or different small and/or macrocells.
The challenge of handling the huge diversity of LTE frequency bands and modes is one step nearer being solved with Qualcomm's latest smartphone chipset which can cope with 40 different frequencies or modes (although not all at the same time). Since they dominate the smartphone RF market with over 60% market share, expect your next iPhone/Galaxy etc. to have one.
Korea is one place to look for those speeds to be achieved first. KT (formerly Korea Telecom) were demonstrating a combined LTE/Wi-Fi small cell which used a simple app to combine the signal from both radios and achieve over 100Mbps. They've deployed over 6,000 indoor small cells to date (mostly in public areas rather than residential), and plan for 18,000 by mid 2013. These are KT branded but made by Qucell (a Korean vendor) with both residential (8 channel box) and enterprise (16 channel box with Cavium chipset).
[Picture above from KT's stand, showing enterprise 16 channel LTE small cell middle (in white with 2 black antenna), smartphone receiving 100Mbps below and Residential 8 channel LTE femtocell in black. The screen on the left shows the 100Mbps aggregate throughput. These are being installed next public Wi-Fi hotspots like that on the right]
Contela explained their latest LTE product (already in use by SK Telecom) which connects up to 8 remote radio heads using simple UTP twisted copper wires to a controller box for enterprise or tower block deployments. It's much easier/cheaper to install than DAS or solutions requiring higher bandwidth to the radio nodes. Contela also announced a new partnership with Hitachi for a combined enterprise and public access small cell solution.
SK Telecom (KT's competitor) told me they had some 3,000 LTE femtocells and 47,000 3G already deployed today with plans for 50,000 LTE femtocells. The vast majority are 32 channel/75Mbps 100mW also supporting 802.11n Wi-Fi with some 8 channel residential units in the mix. SK Telecom also showed some pretty advanced LTE HetNet solutions called SuperCell being developed jointly by Ericsson and Samsung which aims for 100x capacity at 10x less cost. However, this will require both handset and network upgrades to Release 12 specifications (which don't exist yet). SK Telecom still have a fair number of non-4G subscribers today, with a mix of 30% LTE, 50% 3G CDMA and 20% 2G CDMA although 4G is growing rapidly.
Both Korean operators and vendors confirmed the need for a LTE femtocell gateway as part of the solution, to consolidate and optimise the signalling load which could otherwise overload the core network. Stoke, who supply NTT DoCoMo's LTE security gateways, also emphasised the importance of scalability to handling the huge volume of LTE signalling traffic. However it's not all roses - the Head of KT warned Europeans of the "curse of LTE" which has drained heavy investment for limited additional revenue.
Connecting metrocells with wireless backhaul
There were simply too many backhaul vendors for me to visit during the show. The wide mix of technologies on offer each have their own benefits and drawbacks. The 60GHz millimetre wave technology seems to be getting traction and was visibly on show at Vodafone's stand (see below left). Sub10, who have just released their 3rd generation of their product, are offering it in a variety of colours suggesting that they've sorted the underlying technology and have time left to consider the aesthetics (below right). Siklu, another 60GHz vendor, were also active at the show. Tarana publicly showed their newly revealed Non-Line-of-Sight technology, while many established backhaul vendors continued to build on existing solutions.
[Vodafone's booth had a live 60Ghz link installed above the banner]
[sub10's backhaul comes in a variety of colours to suit every taste]
Adding more intelligence at the edge
Ubiquisys were showing off two new indoor small cell products with great long term potential. The larger GM7 Tri-mode using a TI chip supports 3G/LTE/Wi-Fi and seems to have a well thought through hardware architecture with plenty of future upgrade/configuration options although it should be noted that the LTE software not yet been fully "battle hardened in the field". Daughter RF modules allow any conceivable mix and match of radio configurations and support a total throughput in excess of 300Mbit/s. The smaller GM5 using a Broadcom chip is remotely software configurable to be either 3G or LTE.
[Ubiquisys GM 5 can be either 3G or LTE depending on software load - it's even remotely upgradeable from one to the other]
Both come with an Intel Server that offers a variety of "edge intelligence", such as a caching solution using Seguna that also handles lawful intercept, ad counting, billing aspects. They demonstrated proxy upload, which quickly sends a photo to the small cell in <1 second, which then is forwarded slowly to the cloud in due course giving a much better user experience and use of radio resources. See my mugshot uploaded to UbuntuOne below. Ubiquisys have expanded their sales channels to market from 2 (NEC, NSN) to 4, but aren't saying who.
[Proxy upload demonstration. 1 second to send the photo to the small cell, 10 seconds to filter its way onto the Ubuntu Cloud and become visible on the website for all to see]
Meanwhile NSN were touting their Smart Wi-Fi story, which sits alongside their Flexi-Zone higher power solutions and their reselling of Ubiquisys (having recently won UK operator EE).
Operators testing and measuring quality of experience, not just speed alone
System testers can often provide insight into what the industry is really trying to deliver (and measure itself against). IXIA, who provide end-to-end testing solutions for four out of the top five LTE networks worldwide confirm the growing interest in Carrier Aggregation. They also say the testing isn't about basic protocols, signalling etc. but looking more closely at specific end use cases, such as video streaming, facebook updates etc which reflect common user scenarios.
They highlighted two specific technical areas:
- Layer 1 schedule algorithms, which have enormous impact on the end user experience. They've observed some "pretty weird" behaviour during their testing, but report that the industry has matured and resolved early issues.
- Wi-Fi capability. Where before, perhaps handset vendors hadn't been too concerned about testing Wi-Fi capability thoroughly (treating it more like a tick in the box feature), in the last 12 months they've seen a lot more attention being paid to the Wi-Fi service experience which is much more dependent on the smartphone implementation than for cellular service. They see 802.11ac coming to phones very soon; Wi-Gig 802.11ad may be a bit further out.
Small Cell Planning
Radio planning and analysis tool vendors continue to stress the importance of how accurately new small cells are planned and positioned – they need to be within 50 metres or so of the traffic demand or their value could be reduced. Mentum (now part of Infovista) have developed the 3D modelling aspect of their product over the last couple of releases, and also accept feeds from Twitter which can provide very accurate (GPS based) data, albeit skewed by the subset of users and use cases. Keima, Arieso were also showing some impressive graphical analysis.
Alcatel-Lucent have developed their own specific metrocell RF planning tool which quickly illustrates what percentage of traffic is offloaded from nearby macrocells for both the ideal/predicted sites and the actual/confirmed sites available. No matter how good your Self-Organising Network software is, planners will still need to assess where and when to install public access small cells.
TI and Mindspeed seem to be the only vendors with silicon that concurrently supports 3G and LTE, although these are at demo/lab stage. Broadcom can support either 3G or LTE on the same chip, software upgradeable, but just announced they plan to combine baseband and RF on the same chip which would be a unique feauture. Cavium are actively deployed for LTE in Korea and demonstrated HD live video streaming to 8 devices on their stand. Freescale are ip.access choice for 4G, which is shipping to customers for lab trials. Qualcomm (surprisingly) don't appear to have further public design wins for 3G or LTE small cells yet.
[Cavium live demo of multiple streaming HD videos from an LTE small cell. Note the line of 8 smartphone handsets lined up on top of the bottom right hand box]
[Mindspeed showing off a variety of 3G, LTE and TDD demos on their stand]
The software needed for small cells is of course more than just a protocol stack, and the field of 3rd party software vendors is narrowing down a bit. Ubiquisys and Alcatel-Lucent will both licence their software to any potential small cell ODM and a few other players have their own software. Otherwise it's Radisys, Aricent and Node-H that I see making the running – all having both 3G and LTE software solutions. Aricent were demonstrating both 3G and LTE on their stand using a variety of chip vendors, while Radisys (who showed similar demos last year) have broadened their portfolio to include other parts of the network. Both have the resouces to help tailor/adapt/extend their base software. Node-H seem to be more closely linked to Broadcom and take a stricter view of configuration control, being keen to retain a single codebase.
[Aricent demos of both 3G and 4G software running on a variety of different chip vendors]
Symmetricom launched a new edge master clock which provides timing (sync and phase) closer to the small cells. Typically deployed at an aggregation point, this would be more important for areas with poor or no GPS coverage (e.g. inside tower blocks) and where phase sync is more critical (e.g. LTE TDD mode, some LTE-Advanced features). Their small cell sync eco-system continues to grow, adding Pletronics as a second oscillator vendor to Rakon, and small cell vendors CS and Contela.
My apologies for the long length of the report and also for omitting many newsworthy items. It reflects the larger size of the small cell market scope, the number of vendors and activity compared to previous years. It's an exciting topic to cover and I expect to be reporting deeper insights into many of these areas in the next few months.
- MWC reported attendance was 72,000 (rather than 80,000 mentioned above), up from 64,000 in 2012. They recognise the transport issues and promise to improve for next year.