Earlier this week, I spent two days at the LTE World Summit in Amsterdam. More than 2000 great & good members of the telecom industry, including a ton of operators and most of the major vendors. Multiple streams of presentations, a decent-sized exhibition show floor and “big conference” production rather than a small meeting room in a hotel. I was hosting an analyst roundtable on Voice, VoLTE and the Future of Communications, and also presenting a 30 minute presentation on similar topics.
Those of you following my Twitter stream (@disruptivedean) will have seen a fair amountof ongoing commentary, but I thought a few issues were worth drilling into here. I'll be writing a separate post about the Future of Voice, and my upcoming workshops with Martin Geddes, so I won't overdo the VoLTE analysis here.
Overall though, I’ve come away rather pessimistic, despite all the bombastic hyperbole I’ve heard. I’m hearing the same old stories I heard at last year's event – and a lot of them are getting worse rather than better. Loads of hoary old clichés about peak rates and “exponential” data growth. How flatrate plans don’t cut it, long after most of them have been phased out anyway. A ton of unrealistic vendor hype about application-specific policy and charging “business models”.
The big story was also much the same as last year, albeit stated a bit more loudly rather than just implied – there are too many spectrum bands for LTE. At least 8 “core” bands, and another 10-20 also being deployed or considered. Europe will probably get by with three mains ones – 800, 1800 and 2600MHz. With perhaps a little bit of 2100. And some use of bits of TDD spectrum knocking around. Then there’s a variety of US bands, Japanese-specific ones, Chinese ones and a variety waiting in the wings to get approved.
That is *much* worse than 3G, which had one core band for much of the world (2100MHz) and still took a long time to get either coverage or good handset performance.
Bottom line is that LTE spectrum fragmentation is not going to go away. This has a number of implications – firstly, roaming is going to be a real pain when moving “off-net” beyond a single operator’s OpCos, or between regions of the world. In all likelihood, HSPA will continue be used for roaming in a lot of cases. Secondly, handset vendors will likely have to create either regional versions of handset hardware platforms, or make “world phones” that suffer from coverage issues in some markets. Either way, scale economies will be lower, prices higher, testing more problematic and time-to-market longer.
It will not be possible, for example, to have one iPhone variant that supports 3 European FDD bands, Verizon and AT&T 700MHz, the Chinese LTE-TDD variant, something for Japan, and perhaps another US band like AWS or LightSquared. I reckon that Apple will need to create three, possibly four distinct versions of future LTE iPhones.
Now Apple can afford to do that - it only has a single model introduced at a time, it sells in high volumes per device model/version and makes a huge margin on each. In other words, even if each "spin" costs an extra $100m to develop, it's still a drop in the ocean. If it creates three versions and sells 10m of each, it will probably make $2-3bn gross margin on each variant, so it can "wear" the extra hardware development and test cost quite easily.
But it would get very painful for lower-volume devices, or manufacturers that have broad ranges of devices. This in turn means it's probably going to be painful for operators with unusual spectrum bands (eg LightSquared) to get a decent range of decent handsets.
In Amsterdam, we heard repeated pleading from operators - even DoCoMo - essentially saying "Support *my* band! Please! It's really good, and we can get economies of scale & support from all the vendors!".
There are going to be some disappointed players left standing in this game of musical frequency chairs. And everyone else is likely to feel the knock-on effects of component suppliers' hesitation and uncertainty. Some operators will likely hold off on LTE decisions until the spectrum situation becomes a bit clearer.
One other option for LTE that got a little exposure - but was obviously still highly contentious - was that of wholesale-only shared networks like Yota (and LightSquared and a couple of others). I think that although that model makes sense in terms of spectrum usage efficiency, it also poses a risk for incumbent operators that will start to lose control over their core business enabler (the network) and may face a future where all differentiation comes in terms of the (often mythical, and always competitive) "services" layer.
I'll be writing more about the threat from "under the floor" players in the coming months - and why shared/outsourced/structurally-separate mobile infrastructure plays are both inevitable and highly disruptive. I'll be at the network-sharing conference in London next week as well.
One interesting angle on voice and VoLTE that is starting to bubble up - and which I've been suggesting / advocating for some time is that of dual-radio phones. We already see dual-radio CDMA/LTE phones for Verizon and Metro PCS, which use CDMA for voice and LTE for data. This has a distinct advantage over the proposed "Circuit Switched Fallback" standard, in that an incoming voice call doesn't switch off the LTE data channel. I'm expecting to see the same approach appear for GSM/LTE dual-radio phones, but that is much more complex as (unlike CDMA) both radios will probably need separate SIM cards, or two IMSIs on one card. At least one major vendor was openly discussing this approach - but at the moment the lack of standards about handling this type of device is a concern for operators.
Like VoLGA before, dual-radio "velcro" GSM/LTE is a solution that *works* conceptually very well, but it will be interesting to see if the politics of the standards world - and some entrenched interests wanting to ensure that nothing detracts from VoLTE/IMS's uncontested anointment as top solution - get in its way. My view is that this should be the main backup plan or straight replacement for VoLTE: as telephony revenues start to fall, why would many operators want to invest in a new core network and applications when their existing GSM telephony works perfectly?
In my view, operators should invest their future voice/telephony budget in creating new voice products and playtforms - and do the absolute minimum necessary to get decent "old school" telephony working on LTE smartphones. I think the Velcro (yes I know it's a trademark) approach could free the operators to concentrate on creating new and possibly more valuable voice and VoIP applications - before Skype/Microsoft does it for them.
The last comment in this post is about WiFi and LTE. I've had a few conversations recently about the rising star of WiFi usage for offload, onload, roaming and other operator use cases. I think that all of these are extremely important.... but I also sense a dangerous level of groupthink around the "telco-isation" of WiFi. There's a host of new standards and solutions that make bolting WiFi onto 3G/4G networks more "seamless" or more controllable.
Those of you with long memories will know that I have an intense suspicion of the word "seamless". It represented all that was wrong with the ill-fated UMA technology. More than four years ago, I wrote what I thought was the requiem of seamlessness. But it's back, it seems. In a nutshell - seams are important. They're boundaries. Sometimes I want to know when I reach a boundary, sometimes I don't. Things change at boundaries - speeds, policies, price, ownership, security, latency and so on. In particular with WiFi, it is absolutely critical to enable a good user experience between choosing between "operator WiFi" and "private WiFi".
I see far too few advocates of the "private WiFi" use cases - there seems to be an asusmption that WiFi access on smartphones will default to being "service"-mode. I think that is a deeply flawed belief, and unless address will come back to haunt some of the new approaches to offload or operator-provisioned WiFi. More to come in later posts, conference presentations and so forth.
A few quick bullet points of "other" interesting items:
- Apparently, TeliaSonera intends to charge extra for VoIP on its LTE network. Good luck with that. Maybe you can start by providing us with a clear legal definition of "voice"? Downloading a spoken poem? Audio telepresence? Skype video with "mute" switched on? Accessing voicemail? Encrypted speech inside HTML streams? If you're a Swedish-speaking telecoms lawyer, you're going to make a lot of money over the next few years....
- Verizon was being very coy about its rollout and recent outage. Its conference speaker was not even from Verizon Wireless but from the EMEA arm of the company which is mostly the former MCI/WorldCom enterprise services division. Unsurprisingly, probing questions about the progress of VoLTE testing were not especially illuminating.
- Apparently, SMS over the SG interface *is* working. Just that vendors haven't bothered to tell anyone about it as it's not considered sexy. Let's see how the full SMS-over-LTE experience works on future phones though.
- It was good to hear an anecdote from T-Mobile Netherlands that the biggest problem isn't "tonnage" of data traffic, but simultaneous signalling from lots of smartphones and apps in the same place. More interesting still was the massive explosion of the SMS-replacing "WhatsApp" service in Holland, which apparently got to 70% penetration (of smartphones I assume) in just 3 months. Hence KPN's profit warning a couple of weeks ago. (It's worth noting that Netherlands is slightly unusual when it comes to messaging, as it's historically been a low-Facebook use country, instead using its own local social network Hyves)