| It's quite rare for me to take direct pot-shots at other specific analysts. While I'm often confrontational, I try to avoid ad-hominem attacks, or cast doubts on specific items of work. I'll make an exception in this case, because I know the analyst quite well & have a lot of respect for his other commentary, as well as our banter on Twitter, via blog comments, or over a beer at various conferences. I reckon Gabriel Brown (@GabeUK) has called it wrongly on VoLTE and IMS in his latest article . I vehemently disagree - based on discussions with operators, vendors and my own analysis - that IMS platforms are likely to become the main long-term platform for telephony or other applications in mobile operators in the long term. IMS will most likely be used patchily, by some operators, for some purposes, in some places. (Note: I haven't read his full report, which the article is a trailer for, so there may be some more contrarian viewpoints contained in it). I certainly don't see that VoLTE will be the catalyst to " set the stage for a more fundamental transition to operator-provided real-time, rich-media services " - that is more GSMA-style wishful-thinking, rather than a probable outcome. And I completely dispute his assertion that with VoLTE/IMS " investment won't be sunk into interim measures that close off that route to innovation ". What he doesn't discuss is that VoLTE is mis-named. It's actually more correctly called ToLTE - Telephony over LTE, not a generic voice platform. It isn't designed as a ground-up answer to the question "what should voice communications look like in future?". I wrote a report in 2007 saying that there needed to be a 3GPP standard for "plain old mobile telephone" service for IP-based mobile networks, and that it was urgently required to stave off the fast-improving third party VoIP providers. The belated answer is VoLTE, aka ToLTE. It's now 2011 and the expectation is that the majority of 3GPP operators are unlikely to deploy VoLTE commercially until 2013-2014. Which means it won't gain significant uptake among massmarket users until perhaps 2015-2016. Don't get me wrong, an agreed solution for ToLTE would have been important in 2007-2008, because it would have been ready at the launch of LTE last year. It could have been a good defensive move, adding in all the legacy baggage around telephony such as "supplementary services", roaming and assorted requirements for regulation and control. But the problem with defensive strategies is that they have a timing window, after which they're useless. A missile defense scheme needs to destroy rockets in the launch phase, before they break into multiple warheads and re-enter. A football defender needs to tackle someone before they run past towards an open goal. In many, perhaps most, cases for LTE, it now seems clear that defence will be too late by 2014-2016. For all but the fastest and most value-chain controlling operators or countries, several things will have happened by then: A significant proportion of voice application usage will be "non-telephony" by 2015. This is happening already - voice is being embedded in applications and software as a feature, not a service. It might be using a Skype connection as a baby-monitor, or in-game speech between players, or 10,000 other voice use cases. These are not in any sense "phone calls" and are very poor fits with both VoLTE/ToLTE or today's circuit infrastructure. A further substantial amount of the VALUE of both telephony and voice applications generally will have migrated to "cloud voice" by 2015, embedded into websites or business processes, especially for high-end users who will be first to LTE. VoLTE is not designed as being developer-friendly or "mashable" as a #1 goal, which would have been fine in 2007 but not in 2011. (I remember being the chair of an IMS conference on the day that Facebook launched its HTTP-based IM chat. So it's slightly ironic that this debate is happening the day Facebook seems to have trialled its "call" function.) More regulators will have started to allow portability of mobile numbers to third-party VoIP services, and more companies will support it. Google Voice announced it for the US last week. At the moment, mobile number ranges - especially in Europe - are sacrosanct and only available to network owners and their MVNOs. Lets see how long that lasts, and indeed whether numbering retains its psychological importance against other identifiers. The stickiness from SMS, which keeps people from moving to VoIP for primary telephony today, may start to be eroded, especially among LTE users who are likely to have smartphones. Facebook and BBM (and maybe Twitter) are starting to attack the SMS messaging citadel and it is naive to assume that it will remain impregnable. The voice roaming model will have broken down significantly by 2015. It's already crumbling fast, with regional zero-premium agreements and government intervention. Whilst it's understandable that the operators don't want to hasten its demise, they shouldn't be tying themselves up in time-consuming knots, in desperate attempts to revive it. That ship hasn't sailed yet, but it's pulling up the gang-plank and if you're honest with yourselves, you'll recognise it. The fallacy that the operators "own the social graph" via the handset phonebook will become even more wrong that it is today. As I discuss in my report on IMS/RCS it's a great example of self-delusion that Mark Zuckerberg must be sniggering about on a daily basis. There will likely be various dual-radio / dual-SIM phones capable of running simultaneous LTE data and GSM voice, without horrible battery impacts. They will also likely be capable of VoIP over HSPA+ where it is reliable. VoIP over LTE will be a nice-to-have, not a must-have, for these users. By 2015, the concept of a link between an operator's access business and its services will have fractured. Many service providers will want VoIP services that can extend outside their own access footprint, without "interoperability" concerns. VoLTE is unsuitable as a basis for operators "own-brand OTT" activities. Some of the people at operators already know most of this, even though they're trying to ignore the full ramifications. In general, I find that the strategy officers are much more realistic about the challenges than those at the coal-face of the core network - after all, it's not their jobs that are ultimately threatened, so there's less of an incentive to want to BELIEVE in things like IMS as a saviour. But as well as mine and Gabriel's differences in the vision of voice/telephony - which obviously is subjective and opinion-led - there are some other more hard and concrete issues. He writes " it should be possible to at least match [in VoLTE] the service capability of the existing circuit-switched domain used in 2G and 3G networks. " Capability, yes. But quality? Here, there's an interesting gulf emerging between the "establishment" 3GPP, GSMA and vendors, versus the newcomers such as Skype and Google. The old-school approach to QoS is about packet scheduling, managing latency and delay-sensitive traffic, different classes of service and so on. Get the network to minimise dropped packets, and minimise jitter (variations in latency). The new kids try something different. They assume that there will be problems, and look for work-arounds. Clever buffering techniques, new codecs, packet-loss concealment algorithms, echo-cancellation to deal with weird VoIP side-effects, acoustic wizardry to fool the ear. IMS and VoLTE is about prevention, while 3rd-party VoIP is about cure. It's easy to say that "prevention is better than cure", but if it comes years later, is that really true? Skype acquired Camino, while Google acquired GIPS, specifically to deal with speech processing. The GSMA's IR92 specifications for VoLTE largely ignore this whole area. Speaking to a network vendor yesterday, their view is that the bulk of this responsibility in VoLTE falls to the handset vendors. Yet looking at the Internet companies, my understanding is that their QoE (not QoS) engines involve a complex dance between device and server, watching network quality and adapting in real-time. I don't think VoLTE does that. The application isn't really network- or acoustic-aware. One of IMS's core principles is that applications should be "network agnostic", which is a major point-of-failure for the whole architecture. Yesterday, I did a conference call over WiFi to personal hotspot, and then via HSPA to the Internet, on to Skype, and finally out to circuit via SkypeOut. It wasn't perfect, but it was perfectly usable. Roll that forward 5 years, and Skype-over-LTE should be better than circuit (HD, built-in ambient noise cancellation and so on). Google Voice over LTE in 2015 will probably feature realtime translation between languages using speech recognition in the cloud. Yes, if the cell is congested there might be some problems with QoS. But I'm willing to bet that the overall QoE will be miles ahead of VoLTE's. As I've written before, it's pretty clear that 3GPP and GSMA are attempting to use LTE as an excuse to crowbar-in IMS to the operators who've been recalcitrant in adopting it so far. I can use various analogies - it's like Ferrari making a 2-ton boat anchor a mandatory accessory on a new car, or as I put in my most popular-ever blog post, like using LTE as the perch to which to nail the dead parrot of IMS. In some instances, they may succeed - but perhaps to a bare minimum . For instance, I've recently met an operator who's told me, resignedly, that they might put in a small IMS to handle roaming LTE users. However, he has no intention of offering IMS-based services to his own users. To be fair to Gabriel, he does sign off the article with a doubt " Can operators reinvent and extend the voice model into billable rich media services? I think it's too early to say". That encapsulates our differences - I also believe that reinventing the voice model is absolutely something that operators need to do. I just think that VoLTE is absolutely the wrong way to approach it. As for "billable rich media services", the answer is maybe - but only if they ditch IMS as a flawed legacy technology and look for other solutions, even if they are proprietary and don't have the false comfort-blanket of interoperablity. The bottom line is that operators considering investing in VoLTE or IMS need to think twice. What is the actual problem you are attempting to solve, and the business case associated with it? Is it purely a defensive move, and if so is it already too late? Is your "voice" (ie telephony) business as valuable as you think , or is it being inflated by accounting oddities around subsidy repayments and classification of "line access" as voice revenue? Do you really understand what the difference is between Telephony, and Voice as a whole? Is the old-style telephony service worth replicating in LTE at all - will it be monetisable in 2015 and beyond? Will VoLTE actually deliver good-quality speech when acoustic factors are involved too? Are there ever going to be any IMS mobile services that actually add value and provide a good user experience? Or should LTE be used to introduce a proper vision of "services", "cloud" and "voice", with low-margin legacy phone calls kept for the 2G circuit networks that will be here for another 20 years anyway? For more detailed analysis and advice on these matters, please contact me at information AT disruptive-analysis DOT com. |
