Original URL: http://www.theregister.co.uk/2011/06/26/femtocell_congress/

Femtocells at tipping point: Don't want to become also-RANs

From the Macro, to the femto, to the metro, to the micro...

By Wireless Watch

Posted in Networks, 26th June 2011 08:00 GMT

The femtocell industry gathered for its fourth annual world summit this week, and it was clear that significant progress had been made since the last London-based gathering. In June 2010, the tiny base stations had achieved wide acceptance as carrier controlled devices to improve indoor coverage and support offload.

However, actual wide scale deployments remained limited; the talk of new applications and services enabled by femtos had not really borne commercial fruit, and the idea of stretching femto concepts into other areas such as public access metrozones was just that, an idea.

One year on – following a Mobile World Congress in February whose dominant new theme was the ultra-small 3G or 4G cell – the "grown-up femtocell", or metrocell, was also the talk of the London conference, and traditional wireless infrastructure players including Texas Instruments were there with product plans to back up the words.

As they seek to prevent the outdoor RAN falling into the hands of femto specialists, those early movers – firms like Picochip and Ubiquisys – are starting to announce public access products too, bringing a new set of economics to the carrier network.

And in their established territory, the indoor femto is close to the tipping point that should make it a truly mass market product from 2012, while attention has refocused on apps again, as emphasized by Intel's maiden appearance at the show, with its notion of an "edge cloud" of tiny base stations containing significant apps horse-power.

The rise of the metrocell

The Femto Forum announced that femtocells now outnumber conventional 3G base stations, marking the acceptance that small cells have entered the mainstream as indoor devices.

According to research by Informa, there are now 2.3m 3G femtocells deployed worldwide, compared to 1.6 million 3G macrocells and microcells. Operator roll-outs were up by 60 per cent in the last quarter, the fastest growth rate seen to date, numbering 31 deployments. Commercial support is seen at eight of the top 10 global cellcos by revenue.

The research group forecasts that 48 million femtocell access points will be in use by 2014 as the technology expands into the enterprise and the public network. Another interesting result is that 60 per cent of operators surveyed responded that small cells would be more important than macrocells in LTE.

Dr Shahram Niri, director of global LTE/SAE strategy and solutions at NEC, believes the shrinking of cells is inevitable, and less of a culture shock to carriers than some suspect. Indeed, it is already underway with 3G because the only way to increase capacity has been to reuse spectrum.

"Even today, 3G cells are down to a few hundred meters or even a hundred meters in some cases, when it was supposed to be kilometers," he said. "That's because we're learned that of all the techniques we've come up with to increase capacity and the spectral efficiency, the biggest gains we're had was from reusing the frequency, which means we can reuse the same frequency everywhere, but that means smaller cells."

While some vendors and carriers are focusing on femtocells and picocells to infill macrocells to boost capacity, Niri believes that, especially where operators have high frequency spectrum such as 2.6GHz, the best route will be to do LTE with small cells only.

"We're telling customers to not to do macro any more," he said. The new architectures: Metrocells for the public access network have been a key R&D driver for the specialist players for the past year or more.

Picochip makes silicon pay

Picochip, the leading provider of femto silicon, set the ball rolling last fall when it unveiled its picoXcell PC333 system-on-chip, which supports 32 voice/data channels (or 64 in cascade mode) with range of over 2km and handover at 120km per hour.

This indicated the way that consumer grade economics could be applied to a product for the main RAN, but also the additional functionality that would be necessary, such as receive diversity (not always included in residential femtos, but essential to mitigate interference with larger user numbers), and smart signalling.

The metrocell, then, needs to be more robust and complex than its indoor cousin, and will inevitably be far more expensive than the $50 or less targeted for those devices – most carriers are mainly interested in the lower opex associated with features like commodity backhaul and simple installation, rather than in a huge capex reduction.

But it can benefit from the experience in the home market, claim the femto specialists, though there is debate about how far a common architecture is desirable across the different categories.

Picochip's platform is very consistent in supporting residential, enterprise and outdoor base stations with pin and code compatibility, and CTO Doug Pulley says the metrocells use so many of the elements of their residential cousins that they significantly reduce cost and support strong scalability.

"Some day, all base stations will be made like this," Pulley said. "Femto technology will soon be in every base station. With the PC333 we have extended the parameters of femtocell performance to levels that would traditionally have been considered as 'picocell' or even 'microcell'. This high performance coupled with zero-touch provisioning means carriers can routinely deploy femtocells as part of their wide area network roll-outs."

However, others are less interested in the common platform and think the most important crossover from the residential market lies in software or management techniques.

Ubiquisys, for instance, has chosen Texas Instruments' newly unveiled small cell SoCs for its public access product, while relying on Broadcom (via its acquisition of Percello) for its residential offerings.

Both chip providers have steered clear of spanning as wide a range as that of Picochip. TI has said in the past that it could not make the economics of the low-end femto chip work, and now confirms that it will not drill down into the home space but will extend its existing base station architecture for very small cells.

By contrast, Broadcom has pulled off its usual trick of perfect timing in its key market, the home, but is not making many noises about the main RAN.

It bought Israeli specialist Percello last year, giving the start-up the scale to go after major customers, and appears to have made its move just as the sector is set to gain critical mass – a process it carried off most effectively when it entered consumer Wi-Fi.

For Ubiquisys, the key has been to transfer many of the software features of its smaller products to its outdoor offering, including its SON techniques. It announced this small cell a few weeks ago, along with alliances with TI for the baseband and Intel for the apps processor.

But at the summit it was focusing on Broadcom again, launching a fully integrated femtocell home gateway, which will allow operators to deploy multiple services such as fixed broad-band, mobile and entertainment applications from one device in the home.

TI's SoC launch

The UK firm is the first announced customer for the new TI SoCs, which made their official debut this week. The two offerings are designed to be compatible with its existing base station products, but targeted at picocells and metrocells as well as the enterprise.

Brian Glinsman, general manager of communications infrastructure at TI's DSP systems group, believes that tiny, self-organising base stations will grow to about 20 per cent of the total sector in three to five years, from virtually nothing now.

The firm's initial products for that hoped-for surge are the TCI6612 and TCI6614 SoCs, both with production-ready software. Both chips will sample in the third quarter of this year.

The SoCs integrate a mixture of different processing elements, according to the company, including radio accelerators, network and security coprocessors, DSPs and an ARM processor. They support layers 1, 2 and 3 and transport processing for small cell base stations. Both use TI's C667x DSP family based on its KeyStone mul-ticore architecture, which includes both fixed and floating point ca-pabilities in a multicore DSP.

The TCI6612, which has two C66x DSP cores and an ARM Cortex A8, supports 64 users but only one air interface at a time – 2G, 3G, HSPA or LTE. The TCI6614, featuring four C66x DSP cores and the Cortex A8, offers simultaneous dual-mode, running two standards at the same time, and can support 128 users together with MIMO antenna arrays.

While TI and Freescale, which announced a small cell SoC at Mobile World Congress, defend their base station territory from newcomers like Picochip, the interesting unknown quantity is Qualcomm.

The company has been active in the standards and forum, and has announced its own product, but outside the CDMA world this has seen surprisingly little headway to date.

One of its key decisions will be whether to re-enter the base station side of the market, where it only played in CDMA, or to stay focused on seeing off Broadcom in the two US companies' increasingly bloody battle-ground in the digital home.

Other chip firms will be trying their hand too – analyst Stephane Teral of Infonetics told EETimes he would also pick out Altera, DesignArt, Mindspeed and Xilinx. Intel and femto applications.

Intel raises profile

And then there's Intel, the surprise candidate to leverage femtocells to push its silicon into a new space. The company's communications infrastructure division has been raising its profile recently, raising expectations that it will look around for a base station base-band/SoC provider, so that it can assemble a full offering as it is seeking to do in devices.

In the meantime, though, it is playing with its existing products, its Atom and Core apps processors and its network processors. It is taking a two-pronged approach to the new-look wireless network.

One is Cloud-RAN, where it has a key project with China Mobile, working on a fully virtualised baseband processing system in the cloud, supporting thousands of stripped-down base stations, many of which will be small cells.

The other is the "edge cloud", which takes the opposite route, reducing the amount of traffic to the center by putting intelligence and processing power right into the base station. This is a way to get Atom into the base station, as seen in the alliance with Ubiquisys.

One side-effect of that will be to support localised applications for selected areas or user groups, turning the base station into a computer and an apps platform. This has been a goal of residential femtocells, and while those efforts had gone rather quiet at last year's Summit, this year they were firmly back on the agenda.

In particular, the revived interest was highlighted by a collaboration between Telecom Italia, Alcatel-Lucent and Accenture. The partners announced the industry‟s first standards-based software development kit (SDK) for the creation of femtocell applications, harnessing the Femto Forum's API Specification initiative. Initially, the SDK tar-gets Android phones and supports standard Java APIs.

Also on the software side, Continuous Computing announced Trillium Femtotality, a portable software suite which includes turnkey femtocell application software. This aims to help network equipment providers to create femto access points with lower cost and time to market, even if they are not traditional players in the mobile space.

Femtotality comes pre-integrated with Continuous's Trillium 3G and LTE protocol software, and includes radio resource management and SON (self-organising network) capabilities along with OA&M (operations, administration and management) supporting the TR-069 and TR-196 standards.

Copyright © 2011, Wireless Watch

Wireless Watch is published by Rethink Research, a London-based IT publishing and consulting firm. This weekly newsletter delivers in-depth analysis and market research of mobile and wireless for business. Subscription details are here.