By David Chambers of ThinkFemtocell
Since we last spoke with Manish almost a year ago, Continuous Computing have made further strides to increase their customer base for Trillium software stacks used to build femtocells. In this interview, Manish, who is currently their VP of Product Line Management, observes the changing eco-system of femtocell vendors and shares some insights into the rapidly evolving LTE femtocell development community.
Although this interview was held prior to the recent announcement that Radisys will buy Continuous Computing, we feel it is still appropriate to publish it. We’ll be following up the implications for the new business once the deal has completed and their market strategy has been revised. It’s worth noting however, that the new CEO of Radisys will be Mike Dagenais, the current Continuous Computing CEO.
So what’s the most significant achievement for Continuous Computing this year?
We’ve been selected as software partners by 10 “giants” in the silicon ecosystem – all the big names are there including Qualcomm, Picochip, TI, Freescale, Mindspeed and Design Arts. With these bigger players entering the femtocell market, it reinforces the strong momentum growing for the femtocell and small cell concept. It’s all very positive.
Our Trillium software works on all these platforms – we demonstrated this with four different platforms at Mobile World Congress, two each for 3G and LTE – giving our customers the confidence they wouldn’t be locked in to their first choice of silicon vendor.
Does that mean you foresee many femtocell vendors changing horses/silicon vendors?
Not necessarily. I’m sure Picochip, who dominate the market today will continue to work hard to retain their market share. I’d rather say that this validates the femtocell market overall.
We can also see different chipset vendors positioning themselves to target different market segments. For example, Picochip have proven cost effective designs that meet the residential femtocell requirements while TI address the higher power/capacity needs of metro and picocells.
Last year, you weren’t able to announce any LTE femtocell deals. Have you any further progress to report?
We now have 29 3G femtocell customer wins and 21 LTE wins. There is zero overlap – none of our 3G femtocell customers have yet bought the LTE stack. Many have been publicly announced in press releases shown on our website.
With this large customer base, we see ourselves as the “unsung heroes” of the femto industry.
But how many of these large number of customers have commercially deployed their femtocell products?
We are constrained by our customer confidentiality agreements – I’d really love to be able to say more about this – but let’s be clear that we do have a number of customers in Asia and Middle East who are in live network trials with 3G femtocells based on our software today. Additionally, one of the tier 1 femtocell solutions uses our stack in their femtocell gateway. We are working hard to be able to publish more.
The majority of the new LTE customers are for picocell and metro-femto designs. We see LTE being deployed with macrocells first, then outdoor metro-femto and picocells following. Combined 3G/LTE devices could be trialled during 2nd half of 2012 with commercial deployment following on during 2013.
One of the enablers for additional femtocell vendors to enter the market is the practical deployment and interworking of the Iu-h standard. What’s the status today and where is it heading?
There has been tremendous progress to date. We’ve held several IOTs (interoperability test sessions) supervised by the Femto Forum and managed through ETSI. The third is to be held in June at the France Telecom/Orange test labs in France. Several gateways and access points have been proven to support the standard interface.
The driving force behind this standard is the operators themselves – they want to use it to drive costs down and increase their choice and availability of product.
As the market picks up, I expect the same path to be taken as happened with other consumer electronics – the high volume manufacturers (ODMs) in Taiwan and China will dominate in the longer term.
But aren’t there really only a very small number of end-to-end femtocell solution suppliers that hold the key?
It’s true that a few companies have the market sown up and I don’t see that changing. Many network operators have brought in SIs (System Integrators) to manage the services aspects of their business, and this trend will continue. In the future, some of these SIs may source their femtocells from ODMs directly.
I do believe in one certainty though. There are going to be some dislocations in the femtocell market as it continues to evolve. The high volume and need for very low costs means that today’s long “food chain” is unsustainable.
Moving across to LTE now, what progress is being made there?
The early commercial LTE networks were based on Release 8, which was quite a mature standard for a first release. Release 9 made some further improvements and Release 10 has just been published – Continuous Computing is currently working on that and will release a version later this year.
Unlike 3G, the first LTE standard had a femtocell perspective from the outset. But as with any new standard, we are learning from real life field experience some of the issues that weren’t foreseen at the specification stage.
For example, there needs to be some improved interference mitigation between macro and femto – there can be cases where the macro packet data channel can blank out the femto.
Another issue relates to the high densities of pico/metro connected to an EPC (Evolved Packet Core) which can cause congestion – it’s not the data volume that’s the issue, instead the very high signalling load. We worked around that issue by developing our Trillium Fast Path SCTP solution which can scale to very high traffic loads.
And the promise of completely Self Optimising/Organising Networks?
There are still some concerns around mitigation techniques. Some operators also have valid concerns that their macro network may be adversely degraded by small cells. However, a recent Femto Forum report on the subject showed that macro performance actually improves when femtocells are deployed, typically because they take away some of the cell edge issues which would otherwise consume a lot of the macrocell’s own resources.
SON overall is not yet mature at all. The concept has a different meaning and scope to different operators. Does it mean self–optimising or self-organising or even self-healing?
Is there any alternative to avoid interference with small cells?
One strategy which I have been suggesting for some time is to adopt a small chunk of TDD spectrum exclusively for femtocell use, complementing the FDD mode for macrocell use. The TDD option for LTE is proven and built in to the standard mobile device chipsets. It shares much of the FDD protocols and standards, but doesn’t require expensive paired spectrum – it can fit wherever there is a suitable gap.
For many operators, we believe this strategic spectrum investment might well be a better option than overbuilding expensive macrocells.
What’s next?
We see exciting times ahead for Continuous Computing. We’ve had tremendous success and grown impressive market share in femtocell software. We are seeing growing momentum in LTE design wins, and by being deployed in the field we are getting early feedback of the real world issues required to optimise and improve our solution.
