Having witnessed the 4GWorld conference in Chicago last week and some of the reporting from it, I sometimes wonder if I attended the same event. The conference was quite a sizeable affair with something like 8000 attendees. This annual event organised by the Yankee Group attracts mainly North American participants and so has a distinctly US focus. Whilst there was an emerging markets stream, and a presenter from Australia, Verizon were noticeably missing from the lineup of keynote presenters. The fine line balancing WiMAX and LTE camps was carefully tread by the organisers.

Some clear statements on 4G femtocells

• WiMAX is real and here to stay, although LTE is likely to dominate long term
• Domestic LTE femtocells not required for a long time
• Small cells are essential to increase capacity
• US cable operators are all taking radically different approaches for wireless service provision

WiMAX is real

Clearwire were strongly promoting their WiMAX network and gave an excellent demo showing two iPhones side-by-side on a car dashboard. One connected by 3G ran at about 1Mbit/s whilst the other connected via WiFi/WiMAX ran at 4Mbit/s. The faster response to various queries was clearly noticeable, although even streaming video worked reasonably well on 3G.

LTE is still in its infancy, with the first release of standards only 6 months old. We can expect commercial trials towards the end of 2010 with operators in the US and Japan likely to be amongst the first to launch. Analysts predict few devices available in the early days, the first will be USB data dongles.

Overall, the conference suggested that WiMAX had some 2 year lead over LTE, but that heavy backing by the major global operators will mean LTE will dominate in the medium/long term - perhaps with 85% market share. Naturally, the WiMAX camp would disagree, highlighting the lower IPR costs (no patent licenses to pay for).

Answering a spiked question, Bill Morrow CEO said that like any sensible business, he would never rule out selecting LTE technology in the future. This was reported on some websites as though it were just a trivial software update, and a serious probability. Whilst I do agree that LTE is likely to dominate in the medium/long term, WiMAX does seem to have created enough market share to ensure it is sustainable. An analyst forecast of 85% LTE to 15% WiMAX worldwide market share of subscribers in the long term was given.

Data capacity is causing real pain 

ATT shared some of their current pains. Data traffic has grown 5000% on their network in 3 years. This isn't just due to the 8 million iPhones, but also many Blackberries, Netbooks and USB dongles.

Capacity is being addressed by deploying high speed Ethernet fiber connections to the 3G cellsites, installing additional cellsites and upgrading the 3G technology to higher speed HSPA.

With a nationwide footprint of WiFi hotspots, it was explained that some 30% or more of iPhone data traffic is already being offloaded to WiFi.

ATT Plans for LTE 

Having bought some of the 700MHz range of spectrum freed up by the switch off of analogue TV, this will cover some 87% of the US population. Trials of LTE are planned in 2010 with commercial launch of data service in 2011. Voice services will follow in 2012 and will use the IMS core network architecture.

Relatively few device types are expected on the market in the early years, which means data connections will dominate its early use.

Year	Number of different LTE device types
2011	19
2012	34
2013	80
2014	173
2015	287
2016	449

Many questions were asked and answered:

• ATT have no plans yet for LTE-TDD
• No timeframe or decision yet to deploy HSPA+
• Couldn't comment on pricing pans or strategies for LTE
• LTE testing will use both 5+5MHz and 10+10MHz spectrum banding
• Voice over LTE will focus on the IMS architecture.

Some femtocell specific questions too 

With heavy hints in the press that ATT Femtocell launch is imminent, there was still no stated official launch date. No new launch information was released during the conference.

• 3G femtocells are still in trials, optimizing the customer experience. There are also some location/E911 issues being worked on.

• 3G Microcell (femtocell) will be appropriate for 100% of users because LTE handsets will be backward compatible. Data speed will be limited by the domestic broadband, so no urgent need for LTE femtocells.

• Want to see 3G femtocell rolled out quickly and will make sure it works really well on ATT fixed broadband DSL lines.

Cable operators take different approaches to cellular

The cable networks in the US don't just supply Cable TV broadcast channels, they also provide the majority of broadband internet and a growing proportion of business telephony (using VoIP). More recently, they have been taking on the traditional cellular networks in a variety of means.

Comcast and Time-Warner Cable are strong investors in Clearwire , the US WiMAX network. In addition to $1Bn share ownership, they also provide sites for basestation transmitters and high speed Ethernet connections. Dave Williams of Comcast highlighted the benefits of small cells and said they are looking at the WiFi/Femto mix, but was at pains to point out they had made no decision to deploy femtocells at this stage.

Cox is directly competing with the cellular networks by rolling out its own CDMA basestations from Huawei, which will be LTE ready.

Cablevision are backing WiFi instead , with a view to blanketing wide areas with WiFi connectivity.

I saw Juni exhibit a WiMAX 2.5GHz femtocell. This South Korean manufacturer offers one of the few WiMAX femtocells available today.

A lack of Voice over 4G 

The theme of the conference was that 4G (whether WiMAX or LTE) was all about data and not about voice. Sprint clearly stated their view that WiMAX was sold for data only, and they would expect their customers to continue using 2G/3G for voice .

ATT also looked at LTE as providing data capacity to relieve their overloaded 3G network. 

Whilst there are several different architectures to support voice over LTE, there was surprisingly little attention or debate about it at the event. With the long term view being that IMS will take over, early deployments could use:

• the existing 2G/3G network (known as CS Fallback). Verizon have said they'll use this for their LTE launch
• VoLGA. Only appropriate for GSM/UMTS networks, this approach adapts a scheme used for cellular voice over broadband IP (known as UMS or Generic Access Network) to deliver existing 2G/3G core network services directly over IP.
• SIP end-to-end. Already demonstrated and promoted by Mavenir and Acme-Packet amongst others, this approach uses IMS for 4G and interworks with existing 2G/3G core network through smart protocol convertors (known as a convergence server). This architecture also supports the new Rich Communications Suite (RCS) features such as instant messaging and presence between users across different networks.
• IMS. IP Multimedia Subsystem, which requires SIP capable handsets as well as the new IMS core network Billed as the long term objective, IMS can also be used for calls on fixed broadband networks.

I expect the debate over which architecture to use for Voice over LTE to be substantial over the next 6 months or so, and was surprised not to hear/see much of it during this event.

Sprint

Pioneers in the femtocell market, Sprint had trialled their first femtocell back in November 2005. During this early deployment stage the most difficult aspect has been understanding the reason to do it.

The drivers are:

• Coverage
• Cost Reduction
• Customer Control and convenience
• Conservation

When operators provide more customer control and convenience the operator gives up some control. Customers will expect to be able to switch on/off and move the box wherever they want, just like any other domestic electronic gadget.

They still see some challenges for femtocells:

• Scale: Mass scale is needed to drive the cost down
• Integration: Reduce upfront complexity in launching pico/femto
• Standards: Interoperability and consumer choice
• Solutions: Not viewed as nomadic
• Business model: Not yet fully mature

Cisco believe both WiFi and femtocells will win. They believe there are too many different devices in the home and that consumers want to consolidate. Services which are used when moving in and out of the house need femtocells to handover seamlessly. Cisco gave an example of a Webex session at home - this would continue when leaving the home if using a femtocell but drop if on WiFi because the IP address would change.

Continuous Computing announced their 16^th femtocell design win at the event. They believe femtocells win over WiFi by providing presence, location and context for applications. They claim studies report that some 30% of mobile data is used where WiFi is readily available.

I also hear the panel discuss the difference between a femtocell and picocell (a common question these days). With enterprise femtocells matching the capacity and capability of picocells, the consensus was that a femtocell "didn't require an RF engineer" - instead being fully self configuring and optimising. It seemed a reasonable assessment to me, although I don't think today's RF engineers should have any serious concerns about job downsizing - the macro network still needs a lot of specialist work.

Summary

From a femtocell perspective, 4G domestic femtocells are some way off. 3G ones will serve the market need for some time.

Operators do recognise the need for small cells, whether these are WiFi, femto or pico. Data offload from the macronetwork will be essential to meet the capacity demands of some 20-30x today's levels.

4G networks were predicted tol need at least 2x the number of cellsites present today. This seems conservative to me. Femtocells could be used by the operator as part of that rollout, but as part of the own network rather than in domestic sites.

The battle for core network archiecture for LTE is yet to  begin in earnest - watch this space over the next 6 months. US operators are primarily focussed on delivering high speed/capacity wireless broadband using 4G.

Comments (0)