Your browser is not supported. Please update it.

6 August 2008

Qualcomm aims for driving position in defining ‘post-Shannon’ network

‘The industry will have failed it we deploy and optimize networks in the same way in five years’ time as we do now.’ This was the overall message from Roberto Padovani, CTO of Qualcomm, at the company’s recent analyst meeting, setting out the key priorities for the chip giant’s R&D activities over the coming few years. Some of these are predictable ‘ rapid enhancement of HSPA Evolved and LTE Advanced technologies; super-integrated handset platforms; new interfaces and web services. But the driving theme is the fact that the mobile network is ‘within 5%’ of the theoretical limit of Shannon’s Law, and needs to ‘find new topologies that bring the network closer to the handset’.

The most voguish approach to this challenge is the femtocell, and the ability for network designers to combine macro, pico and femtocells, preferably within the same frequency to maximize efficiency. While small cells are attracting the attention of many big names, Qualcomm has been uncharacteristically opaque about its stance on the technology. When it took an investment in UK-based femto start-up ip.access, there were inevitable rumors that this was the first step to a Qualcomm 3G femto chip, but in fact actual silicon from the vendor may be some way off. Its short term interest, it became clear from a range of sources this week, is in addressing the biggest challenge associated with femto/macro networks, interference management.

Improved dynamic interference management in general is classified as a ‘top three project’ by Padovani, and was even given a prominent position in CEO Paul Jacobs’ overview of key issues for the company ‘ ‘we can do wider channels, or we can bring the device closer to the base station, which helps the uplink. We have a lot of thoughts on interference management in a dense network,’ he said in his keynote.

This intense interest is for two reasons ‘ Qualcomm wants to prove to its own satisfaction that interference management can be addressed sufficiently effectively to make femtocell networks operationally and commercially viable (‘and our standards of proof are demanding’, as one engineer put it); and it wants to contribute significantly to solving the problem, in order to shape the embryonic sector in its own image and drive future standards and deployment norms. A chip would only follow these more generic R&D and standards body activities, and even then would depend on Qualcomm seeing sufficient margin and revenue potential in the technology for its business model. At this point it hopes that the femtocell will have fully evolved as a plug and play device within a self-managing network, that could make heavy use of wireless backhaul, not just the most common current option, the user’s DSL line.

All this suggests heavily that the company ‘ outside its CDMA homeland, where it is already working with vendors like Airvana ‘ will sit out at least the first generation of W-CDMA femto chips, and could choose to pull the strings behind the scenes until the LTE phase. By this stage, it will hope to have accumulated significant IPR and software influence that will enable it to devise a well differentiated offering. And few would doubt that, if the economics and interference issue play out for Qualcomm in a couple of years’ time (which insiders hint is the likely timeline for a first product), this is a market to which it would be well suited. It has a long history of embedding sophisticated intelligence in base stations (remember the Smart Base Station platform, pioneering in its day?). And a deep level of base station intelligence will be required to address the four areas that it believes are vital to reap the full potential of the femtocell for operators ‘ interference management, support for multiple air interfaces, dynamic frequency channel management, and improved access control/inter-femto communication.

Qualcomm’s European head, Andrew Gilbert – no stranger to intelligent base stations himself from his Flarion days ‘ said one of the reasons for the company’s strong interest in ip.access (which uses chips from picoChip) was its strong understanding of the interference issues. For now, such partnerships ‘ geared around sharing knowledge and expertise rather than direct product collaboration ‘ look like being enough for Qualcomm as Padovani’s team steps up its efforts to ensure that the chip designer has a key role in defining how the ‘post-Shannon’ mobile network will look.

Airave:

Despite all the focus on W-CDMA, the first commercial deployment of something approaching a femtocell network has been in CDMA, with Sprint Nextel’s Airave system, made by Samsung.

This product is bulkier and more basic in capability than the femtocells envisaged for 2009 onwards, but offers some of the simpler of the femto’s promised advantages, notably good indoor coverage (still a huge issue in many parts of the US) and the opportunity for homezone tariffs. Having trialled the Airave, to generally strong acclaim, in a few markets, Sprint is now poised to push out the offering on a national basis, at least in metro areas.

This symbolizes the argument that carriers were having last year, about whether femtocells or Wi-Fi access points with dual-mode handsets were better for supporting fixed/mobile convergence, homezones and indoor coverage. The additional promise of femtos has largely convinced advanced carriers that femtos will be the long term way forward, even if Wi-Fi is a useful stepping stone. Support for new applications; a potential shortcut to early LTE roll-outs; the ability to create hotzones and fill in dead spots outside too; self-management combined with strong operator control ‘ all these make a femto network, assuming these benefits can be proven, a more attractive route for 3G and 4G cellcos. But Wi-Fi has enabled some operators to make a move far earlier than will be possible with femtos, and the poster child for this has been T-Mobile USA, whose HotSpot@Home offering is directly targeted by Sprint’s Airave.

However, the charging structure that Sprint has adopted for its nationwide launch next week may be a competitive disadvantage against T-Mobile. Sprint will charge $100 for the device, $5 a month for femto connectivity and then layer calling plans on top (from $10 or $20 for basic family calling plans). This it will make customers pay both for the femtocell and the service, whereas in most Wi-Fi models the device is free or heavily subsidized with a contract. By asking customers to subsidize roll-out, rather than using subsidies itself to accelerate uptake, is bold, and reflects either Sprint’s confidence in the high level of demand for better indoor coverage and other benefits, or a misstep that will need to be rectified soon as T-Mobile also steps up its expansion.

Meanwhile, also on the CDMA side, Airvana is collaborating with Hitachi on a femtocell platform optimized for the Japanese market.

The interference management issue:

Some key questions that are being tackled by the early specialists ‘ and must also be enshrined in standards ‘ include:
‘ Management of overlapping macrocells is well supported in 3GPP standards, but will these mechanisms be sufficient when there could be millions of small cells in close range?
‘ Will current adaptive radio control mechanisms enable the advanced user experience that consumers will demand?
‘ How will the user stay on the femtocell, when within its range (and so gain advantages like homezone tariffs) if the macro coverage is also good? One way is to crank up the power of the femto, but this could also have interference impact for the macro network.

All these issues will get more complex as femtocells proliferate and as users move them around. The early femto vendors have gone a long way to addressing the questions in theory, but need to prove their arguments to operators in real world networks with huge numbers of base stations. The Femto Forum will have an important role in coordinating activities such as large scale modelling and simulation, and arriving eventually at standard approaches that can be submitted to bodies like the 3GPP. This should accelerate progress as work will be less fragmented and architecture-specific, as should the participation of companies with major R&D resources and intelligent base station experience, like Qualcomm.

In the medium term, there will also need to be changes to handset design to maximize the benefits that users gain from femtocells in the real world. Although the miniature base stations work with identical handsets to macro ones, the femto is expected to do more ‘ at least in the second wave of roll-out ‘ than just improve coverage and capacity. It should also drive new applications and new behavioural patterns, especially when using the handset within the home, where the femto and smartphone could become a hub for content and a centralized control point for various home networks and media systems. For these new applications to be fully usable, there will need to be new features built into phones, and handsets will need to be ‘aware’ that they are attached to the femto, and therefore behaving in a home-based manner, or (even more complicated) that they are connected to an outdoor hotzone. While Release 8 of the 3GPP specifications include various femtocell related items that will help improve how devices work with ‘HomeNodeBs’, as 3GPP calls them, these standards will not appear in phones for another two years. The optimization of handset operating systems and control software for femtocells is another area where Qualcomm believes it is likely to be influential.