LTE for heavy mobile broadband users? A paradox

Various presenters at the LTE Summit have said things along the lines of:

- 5% of mobile broadband customers generate 70-80% of traffic (eg the Telenor subscriber downloading 230GB per month over HSPA)
- A few inner-city HSPA cells are overloaded (3 or 4 carriers) while the ones in rural areas are essentially empty
- Most of the traffic comes from PCs or iPhones

But what's certainly not obvious to me is that the 5% of heavy HSPA users suddenly become profitable if you give them LTE instead, especially in expensive new spectrum for 20MHz channels. Why not just give them a femtocell for use at home, implement strict caps, or just get rid of them entirely?

Or that upgrading a few urban "hotspots" to LTE solves all the problems, given the need for everyone to have new LTE devices (you can't really say "only upgrade your dongle if you think you might use it in London W1 some time in the next year")

The only argument for short-term LTE deployment at the moment seems to be around spectrum flexibility, and the option to use thin slivers in refarmed GSM bands, or awkward allocations elsewhere that don't map onto HSPA's 5MHz channels. For most operators, that's unlikely to be an easy business case.

LTE, mobile broadband & SMS - VoLGA vs. IMS

Just watched a presentation on the UMA-derived approach to voice over LTE, called VoLGA, which I wrote about a few months ago at launch.

As well as the discussion about voice, re-use of existing circuit cores vs. IMS and so on, there was a fascinating comment by the speaker (from T-Mobile) about the "under the hood" use of SMS in mobile broadband dongles.

It's used for a wide range of functions - most visibly in enabling texts to be sent by the user from the operator dashboard connection sooftware. But beyond that, it's used for things like roaming notifications, internal configuration of roaming lists and other settings, and assorted others. There are apparently "many more than 10" systems at T-Mobile that rely on SMS in the context of a mobile broadband computing service - and which would need to be changed if SMS was not supported easily.

This all represents a huge issue for early releases of LTE-based mobile broadband. Despite the argument by 3GPP that "IMS will eventually get deployed, and phones will take a while to develop, so voice on LTE can wait a while", it seems likely that SMS will be needed from Day 1 .

In any case, I completely disagree with the 3GPP representative that every LTE network will inevitably feature an IMS back-end. If it becomes almost mandatory (and some of the "hooks" in the radio and EPC bits of the standard are heading that way), it's a good way to ensure that it won't get deployed universally. Many operators are implacably set against IMS, as it still has huge limitations. (For example: would an MVNO on an LTE network need its own IMS, or be forced to use wholesales apps via the host operator's IMS?)

There was also a comment from 3GPP that LTE could use CS-fallback for SMS. How that would work in the context of a 3G dongle and PC I'm not sure - presumably this would mean that an incoming text (or system message) would force the connection to degrade. Fine for an occasional message - but not for a teenager sending & recieving 200 SMS's per day from their laptop. Can you imagine any other type of broadband link having to drop speed each time an email or IM arrived?

I don't think VoLGA is perfect - once again, it will need extra software in the device protocol stacks. But given the complete abdication by the vendor and standards bodies in sorting out voice/SMS over LTE in timely fashion (ie 3 years ago), it's currently the best workaround I've seen. The fact that it's not even been included in 3GPP's release 9 workplan is ludicrous.

[Sidenote for the "voice on LTE can wait" believers: pretty much every laptop user connected via LTE, especially where it's used a fixed broadband substitute, will want some sort of voice capability. Right now, the only options are Skype or various Internet SIP alternatives. Are you really happy to give them an open goal?]

Evolved Packet Core in LTE

Many people in the mobile industry are unaware that LTE, the radio Cinderalla, also has an ugly sister, called EPC (Evolved Packet Core - originally it was called SAE).

Many things about EPC are actually worthy - "flatter" IP network, lower latency, the ability to mesh between base stations, getting rid of a lot of other legacy equipment like SGSNs and GGSNs and so forth.

However, there is one thing that appears to be missing - the option for local offload of traffic. Everything is still backhauled via various gateway and policy boxes. It's not possible to say "that's a 3GB web video download - just short-circuit it straight to the Internet". It still has to go through the core network.

If I'm understanding this correctly, this has a couple of important implications:

- Roaming traffic will still need to be backhauled via the home network, even if it's just plain-vanilla Internet data. You can't just dump it to the web in the visited country. So you can forget about LTE roaming being able to compete with local WiFi (or a locally-acquired LTE SIM) on price, because there's still a huge extra overhead of network elements and unnecessary transport involved.
- In LTE femtocells, the option for "split tunnels", for example to offload web traffic or operator services like IPTV in the home gateway, or DSLAM / cable headend, won't work. Again, the traffic needs to go through the mobile EPC core. You can also forget about using LTE instead of WiFi for UPnP or other types of home networking via a femto - the traffic will need to be "tromboned" in and out.

Are there risks associated with an LTE "monoculture"?

I'm listening to a presentation on LTE-TDD, highlighting the close integration between FDD and TDD branches of the technology.

The underlying assumption in the industry is that it's wonderful that we might have, for once, a single global standard for wireless networks, albeit in two "flavours".

The subtext (or sometimes explicitly stated position) is that alternatives like WiMAX are thus undesirable.

But I'm wondering whether there is a dark side to full standardisation - even with its associated conveniences and scale economies. What happens if we create what biologists call a monoculture? In agriculture, this carries serious dangers.

The most obvious risk is that the impact of any catastrophic risk is magnified hugely. Any vulnerability - either technical, or perhaps IPR-based - would have universal effects if exploited.

In addition, certain business models are hard-coded in, or out, as a result of specific standards. This is already occurring in LTE because of requirements for mandatory SIM use, and because there is no option for local offload of traffic without backhauling via the packet core.

I'm not convinced that the short-term gains from standards and scale economy necessarily outweighs the long-term losses from business model flexibility and reach. An LTE monoculture may result in the eradication of valuable gene-lines elsewhere in today's mobile environment.

It's worth considering that what the technology industry calls "fragmentation", an ecologist would call "biodiversity".

Microsoft - bearer-aware applications are the way forward

I just listened to an ultra-detailed and very informative presentation by a technology planning director for Windows Mobile. Irrespective of the current lagging commercial status of WinMob in today's market, I was hugely impressed with the vision.

In particular, it was the most coherent presentation I've heard about the future split of mobile app functionality between device and server ("the cloud"). Various examples were cited of using LTE as a way to offload processing from phone to network, such as realtime voice and image processing, which would overwhelm most handset's chips and batteries.

There was also a discussion of using the handset as a web server - which given that I heard almost exactly the same thing from Nokia at the Telco 2.0 event last week, suggests to me that it's likely to be a big deal quite soon.

He also put up one of the few slides I've seen at a public conference that acknowledged the important future role of both connection manager and some way of exposing information about radio bearers. It mentioned the magic words for APIs that give "Notification of radio states & options" and "Ability for apps to choose radio options".

This is precisely what I've been talking about for several years, and to be fair, it's also something that fits in with Symbian's Freeway architecture and some of the bearer-aware iPhone apps I've seen.

In Q&A he also agreed with me on femtocell awareness - ie applications ideally being able to distinguish between femto & macro (& WiFi) connections and behave differently. (This was one of the main topics covered in the Disruptive Analysis Femtocell-Aware Handsets report a while back)

At the LTE Summit. With a herd of elephants in the room.

I'm in Berlin at the LTE World Summit for the next couple of days. It's a well-attended event, with a better-than-average mix of operators vs. vendors, as well as broad range of heavy hitters.

I'll try and put up a few posts over the next couple of days, but already a few things are standing out. Never mind "an elephant in the room", I'm wondering if there's an entire herd, plus a few hippos thrown in for good measure.

The two largest and ugliest pachyderms are:

- Voice
- Business model

A panel of FT/Orange, T-Mobile, GSMA, Ericsson and 3 failed singularly to deliver a compelling answer on how we'll get voice working over LTE, especially in the short term. T-Mobile wants to use IMS in the longterm, but reckons it might need UMA/VoLGA in early deployments. Orange is much less convinced about VoLGA (is Uniq *really* that successful?) and thinks 2G/3G fallback is the way forward. Ericsson is still flogging the dead horse that is IMS MMtel (remember that?). The GSMA wants us to focus on data for LTE, and thinks it's actually quite impressive that we're even thinking about voice now, rather than as a post-hoc "oops" after LTE launches. And 3 seems to be skeptical about LTE for the very reason that it's already looking really fragmented for things like voice. (I reckon 3 might spring a surprise with "native Skype" on HSPA+ - or else will stick with GSM for a long time to come).

Up to a point the GSMA is right - at least we're thinking about this now. Although it was already a clear problem 2 years ago, when I first published a report on VoIPo3G. But the notion that it's all on track, especially given the protracted timeframes for handset platform & device development, is untenable.

(Separately - I'd love to be a fly on the wall when Ericsson tries to pitch the idea of an MMtel-enabled LTE iPhone to Apple....)

My view is that voice on LTE is looking ever-more intractable. VoLGA looks like a good idea, but I'm not sure that every operator will want to go through the hassle of deploying all the gateways & tunnelling paraphernalia, although some will. 2G / 3G fallback strikes me as a nightmare, especially for high-end devices, unless it's possible to run multiple radios simultaneously. Ed Candy talked about the issue of time needed to change network - but the other elephant is what happens to any running 3G data services if you want to make a call. Talk on the phone and use Google Maps at the same time? I don't think so, unless you have very clever multitasking connection managers & radio layers. Add into the mix the impact of femto vs. macro voice for LTE, and it gets even messier - the optimum approach could well be different when on a femtocell.

I'm starting to think that my old favourite theme of bearer-aware applications will be essential for voice on LTE. The dialler/telephony app (perhaps in conjunction with the app server in the network, especially for inbound calls) may need to make smart decisions about the best way to "game" the network for phone calls, based on an intelligent view of latency, quality, concurrent applications running and so forth.

The other thing that's going to be problematic is that of business model for data-centric devices. Moray Rumney from Agilent asked an impassioned question to the panel about the ridiculousness of roaming tariffs, and how they can be mitigated (or whether WiFi will remain as the nomadic broadband standard). The GSMA response was equivocal, pointing out how happy the association is at the idea of extra revenue from data roaming, and pointing out that rates are coming down as a result of regulatory intervention.

This is palpable nonsense of the first order - the fact that we've gone from 5 orders-of-magnitude overpricing, to just 1000x, on the pain of regulator pressure, is hardly cause for celebration or smugness that the problem is solved. And in any case, that's only intra-European roaming anyway. Even more egregious was the comment that the hotel's WiFi (fast & free - thanks Informa / Hotel Palace) was costing someone a lot of money, and therefore this highlighted the "value" of the cellular alternative.

Now it's absolutely true that had it not been "free" it would have cost a ludicrous €22 per day at this particular establishment. No surprises there - it's provided by Swisscom, which is always consistently outrageous.

But the delegates are not paying for it. Someone else - Informa, the hotel, or one of the sponsors - is picking up the tab.

This is a business model which simply would never work for LTE, because the underlying architectural standards work against it. The reason that third-party sponsored WiFi works, for a room of 200 people, from maybe 30 different countries - is that it does not require cumbersome roaming mechanisms, or a physical SIM card.

In my view, the lack of a SIM-less option for LTE is a huge mistake. It completely destroys the possibility for a whole raft of innovative session-based, ad-hoc, temporary or location-specific business models for data connectivity. Don't get me wrong - absolutely there should be SIMs for normal subscription services and various value-adds. But there are plenty of other commercial opportunities for data services in particular, for which the need for the user to obtain a physical SIM would be a deal-breaker.

The industry is leaving money on the table by ignoring this - it will instead continue to be picked up by WiFi, fixed networks and WiMAX, all of which are more flexible in their authentication options. For now, this is masked by the rapid increase of monthly-subscription contracts. But the pool of suitable customers is limited, and as the growth tails off, it will be essential to find new pricing models.

I think this will get fixed somehow (or perhaps worked-around with clunky roaming options) - but it needs to be examined ASAP for LTE to be as much of a success as its proponents hope.