SON, your time has come – Julie Bradford, Managing Consultant

While self-organizing network (SON) technology might not inspire the same column inches as the hype surrounding 5G, for example, there are signs it’s finally starting to come back into the limelight.  In fact, I was recently part of a London conference wholly devoted to the subject, where a series of operator presentations left delegates in no doubt about the value and challenges of SON and automation both in today’s networks and on the path to 5G.

With 5G often described as a ‘network of networks’, such dense HetNets can be characterized as ‘multi-x environments’ – multi-technology, multi-domain, multi-spectrum, multi-operator and multi-vendor. As Small Cell Forum recently noted, these networks ‘must be able to automate the reconfiguration of [their] operation to deliver assured service quality across the entire network, and flexible enough to accommodate changing user needs, business goals and subscriber behaviours.’ And in networks comprising macrocells, large numbers of urban, enterprise and residential small cells, Wi-Fi access points, distributed radios and DAS antennas, it’s clear that optimization will be literally impossible without advanced SON at the heart of such automation.

But it’s not just the growing diversity of access mechanisms driving demand for sophisticated automation. The variety and complexity of services that are delivered by both network operators and over-the-top providers imply a huge range of customer experience expectations and network performance. Clearly, voice services over LTE are sensitive to packet loss, gamers struggle with latency, movies-on-the-go demand reliable buffering and so on; this is before the tsunami of 5G offerings ranging from augmented reality and the tactile internet, to sensor monitoring and first responder connectivity.

In addition, a number of conference speakers in London discussed virtualization and SDN in 5G networks.  Here there must surely be an opportunity for SON to help dynamically manage the ‘network slices’ of spectrum, sites, equipment, inter site transmissions etc. to deliver the right combination of services in the right place at the right time.   Many mentioned the idea of SON having to move from being re-active to pro-active as we move forwards on this path.

But in all this there is a danger that we are getting ahead of ourselves. At Real Wireless, part of our job is provide independent guidance around current investment decisions.  Many of the presenters at the conference were encouragingly operators who are actively using SON in their networks today.  However, much of the focus of real deployments was on automating radio planning and maintenance of sites which is still the tip of the iceberg in terms of what SON could achieve.  Also issues around interoperability, scalability and taking SON beyond the RAN appeared to remain a concern for operators, indicating that there is still plenty of work to be done to realize the full potential of SON.

Caroline Gabriel’s recent operator study concluded that automation and SON become entirely critical to the HetNet business case at a density level of about 10 cells per macro/50 cells per square km point, though SON was also considered highly desirable beyond three cells per macro/15 per square km. As network densification becomes more pressing, the case for reliable and interoperable SON will become increasingly urgent.

Real Wireless report on key costs in virtualising small cells

Last week at The Small Cells Summit in London, Small Cell Forum announced their release 5.1 document suite which is the first phase of its work on small cell virtualisation. Alongside the Forum’s own work to assess the technical benefits of the virtualising of small cells in cellular networks, Real Wireless examined the key cost elements in deploying and operating small cells in urban areas that would be most sensitive to a move to a more virtualised network architecture. Core to this assessment was understanding both wireline and wireless transport options to small cell sites and considering how these options reduce and costs potentially increase as architectures move from traditional distributed RAN (DRAN) LTE networks today towards classical cloud RAN (CRAN) architectures with remote radio heads and CPRI interfaces at the network extremes with strict latency requirements in the order of 250us and bandwidths in the order of 2.5Gbps.

image001

Key findings from the study were:

  • Our results challenge the traditional view that dark fibre is prohibitively expensive. On a five-year TCO basis dark fibre costs can be commensurate with managed fibre given recent falls in dark fibre prices.
  • Most transport options today can meet bandwidth and latency requirements up to and including a MAC/PHY split with latency requirements of 2 to 6ms if the centralised processing is done at a local macrocell.
  • The CPRI/ORI case of the most challenging virtualisation split considered is supported only by Sub-6 GHz and dark fibre in 2015 but could be supported by all other transport options by as early as 2020 (except copper). However, we note that most wireless options will only support these requirements over short links and with good line of sight and that managed fibre products do need to evolve from packetised services offered today.
  • Across the transport options surveyed, the cost increase for supporting CPRI split beyond a MAC/PHY split was most dramatic for managed fibre and microwave (assuming these will support CPRI by 2020).
  • Across the transport options surveyed, the lowest five-year TCO for a CPRI split was found for Sub-6 GHz with $32k versus a managed fibre CPRI of $95k in 2020.
  • The virtualisation cost, although very sensitive to whether NFV is done at a macrocell or data centre, is still a very small cost in the bigger scheme of the total TCO of a small cell site and has only a marginal impact compared with the transport connection to the small cell site.
  • The power cost is also very small (in the range of $100/year) and has little impact on the overall TCO and has little variation between transport options.

The full report is available at: http://scf.io/en/documents/158_-_Business_case_elements_for_small_cell_virtualization.php

Will cognitive radio, dynamic spectrum access come of age in 5G?

Around 10 years ago, the Defense Advanced Research Projects Agency (DARPA)’s Next Generation Communications program constructed a prototype cognitive radio system, which utilized dynamic spectrum access for its communications. By identifying unused sections of spectrum in the area it was operating, it was hoped up to 10-times more spectrum would be available for transmissions. This highlighted a growing interest in the defense community in dynamic spectrum access techniques which had been developed with the challenges of battle-space spectrum in mind, but also apparently had applicability in commercial environments in terms of making more efficient use of valuable spectrum resources and potentially leading the way to spectrum trading. The XG program was one of the largest cognitive radio projects at the time but interest in Cognitive radio was by no means limited to the U.S.

Martin Cave’s audit of public sector bands in 2005, which highlighted just how much more efficiently U.K. defense spectrum could be utilized, provoked interest in the topic in the U.K. This was produced alongside Ofcom’s Spectrum Framework Review, which set out ambitious targets for a general move from the traditional “command and control” approach to spectrum licensing to a more dynamic approach based on “market mechanisms” with the overall ambition of realizing better value from spectrum for the U.K.

With the switchover to digital television and release of TV white space, a debate was ignited over whether DSA could be applied to these civilian bands too. The obvious example of this has been the activity around TV white space, although the Federal Communications Commission discussion on 3.5 GHz is also significant.

However, the digital TV switchover was six years ago and the commercial roll out of white space devices is still fairly limited due to the complications of deploying these devices in practice. Concerns over the so-called “hidden node” issue (interference provoked by the failure of one device to detect the presence of all other devices) and how devices with different spectral views would liaise with each other have meant that the regulation of these white space devices has taken some time to agree.

In attempting to overcome these limitations, regulators gradually shied away from a pure spectrum sensing approach, towards the introduction of beacon signals to identify usage, before settling on the use of a centralized database of white spaces in each location that is used in addition to spectrum sensing.

But even then, the practical use of TV white spaces has continued to be fairly limited. Vendors and operators have struggled to find an application that suits the availability of white spaces, as well as handling the lack of guaranteed spectrum.

This same philosophy is being proposed for 3.5 GHz in the U.S., where some locations have other users (e.g. marine radar), but the combination of database and sensing could allow this band to be used. This is especially important as 3.5 GHz is one of the few LTE bands that is supported globally, so there is a clear commercial imperative.

Enter 5G

At the recent 5G Huddle, rethinking how existing technologies make use of spectrum was a key topic of discussion, with spectrum sharing a major part of this.

There are some strong arguments for why this would be sensible:

  • We’re starting to reach the limits of what we can achieve through higher order modulation schemes, with any gains insufficient to keep pace with demand.
  • We may still be making some gains with regards to multiple-input, multiple-output and CoMP, but again, not at the same rate that demand is increasing.
  • Small cells, which are increasing in usage, and network densification, levels of which are also increasing, both lend themselves well to spectrum sharing.
  • –he last variable available to us in our attempts to increase capacity is spectrum, and (at least in theory), DSA maximizes availability and efficiency of spectrum across all operators

On that last point, this is of course only if it is deployed correctly, with polite protocols for communications.

However introducing dynamic spectrum sharing to “5G” would surely result in 5G just suffering from the same technical issues that cognitive radio has encountered before.

After all, one of the key differentiators of cellular over many other wireless technologies, such as Wi-Fi, is the guaranteed quality of service. Indeed, we have previously examined how exclusively licensed spectrum loses value as the sharing arrangements increase uncertainty for operators.

Wouldn’t 5G lose this edge if spectrum access became dynamic and without guarantees?

At present, you would be correct, but it is unlikely anyone would be satisfied introducing such a glaring problem into 5G. Rather the key difference between earlier cognitive radios and 5G is that, as demonstrated with the discussions at the 5G huddle, major commercial vendors and operators are putting significant research time and investment behind the technology.

Perhaps this time around, with the full weight of the industry behind it, and with an appropriate understanding of what operators need from spectrum sharing conditions to offer high-quality services, cognitive radio and DSA can really come of age.

This blog post originally appeared as part of RCR Wireless’s Analyst Angle, where the industry’s leading analysts discuss the hot topics in the wireless industry.

The changing face of the femtocell ecosystem – who is best placed to succeed?

The femtocell industry has traditionally been dominated by start up companies dedicated to building small cell products such as 3 Way Networks, Ubiquisys and ip.access.  However, as wireless broadband traffic forecasts have continued to highlight trends such as the high demand for capacity indoors and a high proportion of traffic being generated by a small proportion of users the case for smaller cells has gathered support.  This was demonstrated at Broadband World Forum last week where operators agreed that capacity bottlenecks won’t be solved by spectrum efficiency improvements of 4G alone and that smaller cells will be needed [1].  Recent auctions of spectrum at both 800MHz and 2.6GHz in Germany have also facilitated plans for a two tier network topology, further fuelling the interest in smaller cells [2].

This growing interest in smaller cells has led to some big industry players joining the traditionally niche and UK centric femtocell ecosystem.  For example in March this year Qualcomm announced that it would be providing its femtocell chipset to ZTE [3].  Most recently Broadcom has entered the market by purchasing the Israeli femtocell chipset vendor Percello [4].  Freescale has also recently joined the Femto Forum and has announced their intention to develop LTE System on a Chip (SoC) platforms suitable for femtocells [5].   This is perhaps an expected development to match the femtocell platform product range already launched by FSL’s rivals TI.


Figure 1 – The femtocell ecosystem with supply links shown

With these changes to the femtocell ecosystem, who is now best placed to succeed in this growing market?  While only time will tell exactly who the big winners and losers of the femtocell industry will be we can make some observations about what has worked well so far such as:

  • Being first doesn’t necessarily guarantee success
  • It’s good to know your niche in the ecosystem and to stick to it
  • Know your target market; home or enterprise

Being first doesn’t necessarily guarantee success – The first UMTS femtocell access point product was produced by Cambridge based start up 3 Way Networks [6].   The product successfully attracted interest from investors and 3 Way Networks was acquired by Airvana in 2007 [7].  However, being first to market unfortunately hasn’t guaranteed success and Airvana recently announced that they are discontinuing their UMTS femtocell product range [8]

It’s good to know your niche in the ecosystem and to stick to it– One of the success stories of the femtocell industry to date is the rise of picoChip.  Established initially as a flexible baseband processing solution for software defined radio platforms, picoChip were quick to pick up on the promise of femtocells and have focused their product roadmap around producing a cost optimised off the shelf femtocell PHY chipset with an onboard processor to allow femtocell access point vendors to add their own higher layer stack and interface to the femtocell gateway.  Making a clear decision to be a chipset vendor and sticking to this vision has also produced dividends for Percello who were recently acquired by Broadcom.

Know your target market; home or enterprise– Traditionally the femtocell industry has focused on the home market.  However, more recently femtocell vendors have turned their attention to the enterprise market which requires providing coverage to more users and over a larger cell radius in an office rather than home environment.   These are two very different markets with different cost considerations and technical challenges.  In the residential femtocell market a low unit cost is key and backhaul restrictions are a big concern.  The enterprise market is not as sensitive to cost and while backhaul may not be as much of a concern in an office environment the number of users and throughput will be.  Vendors trying to serve both markets may well be left with comprises in their product that a more focused vendor hasn’t had to make.

Over the next year it will be interesting to watch how the increasingly congested femtocell ecosystem evolves and whether the smaller vendors with more experience in this sector will be able to navigate their way past the challenges of the better known but relative newcomers to the femtocell industry.

References

[1] European telcos say LTE will not solve capacity crunch, http://www.totaltele.com/view.aspx?ID=459850&G=1&C=4&Page=0

[2] German mega mobile spectrum auction ends, http://realwireless4g.wordpress.com/2010/05/20/german-mega-mobile-spectrum-auction-ends/

[3] Qualcomm Snags First Femto Wins, http://www.lightreading.com/document.asp?doc_id=189633

[4] Broadcom Corporation to Acquire Percello Ltd., http://www.broadcom.com/press/release.php?id=s523205

[5] Freescale plans basestation-on-chip, http://www.eetimes.com/electronics-news/4200872/Freescale-plans-basestation-on-chip-

[6] World’s First Commercially Available 3G Femto Cell, http://www.3g.co.uk/PR/Feb2007/4221.htm

[7 ] Airvana Acquires 3Way Networks, http://www.airvanacom.com/products/news_378.htm

[8] Airvana’s Femtocell Market Focus, http://blog.airvana.com/airvana-blog/2010/09/airvanas-femtocell-market-focus.html