Inside-out deployment of LTE using femtocells

There's been a lot said and written recently about the concept of rolling out LTE in homes or hotspots first with femtocells, rather than deploying a full macro network.

I've been calling this an "inside-out network" approach.

On the face of it, I can see a lot of positives - it potentially reduces the overall capex requirement needed for network rollout, and solves a lot of issues about indoor coverage for markets where LTE is most likely to be deployed in the 2.6GHz band.

But I'm a bit wary about some of the assumptions being made. Particularly comments like "In 2013, 60% of mobile data usage will be indoors".

Maybe. Maybe not. It's a brave person who'll pre-judge what applications will be used, on what devices, in what contexts, five years out. A year or so ago, nobody was expecting the use of Google Maps on handsets to be one of the prime drivers of macrocellular 3G traffic. Although I'm perennially skeptical, maybe someone will have finally worked out a way to get people to consume mobile TV. Nokia's been talking up the idea of a sort of augmented-reality overlay, superimposing extra information on a view of the "real world". Maybe we'll all have Bluetooth head-up displays, showing streamed video adverts on what we'd normally see as blank walls.

Are operators going to be happy about possibly being held hostage to future application innovation? Someone comes up with a great revenue-earning new service - but it's used outdoors, so it can't be deployed on an inside-out network.

And in any case.... assuming that a lot of that mobile data use is indeed indoors, what % will be on devices that also have WiFi in them? PCs, iPhone, most high-end smartphones (Blackberry Storm excluded, obviously...). Over 90%, perhaps 99% of that indoor data could be offloaded to WiFi.

And I certainly don't believe all the femto hype about substituting for WiFi, especially in markets with 4-5 competing LTE operators and no national roaming, so you'd need an array of separate operator-specific femtos. Yes, there might be the odd single-person household or combined family-plan home that could work for, but they'll be the exceptions, not the rule. And you'd definitely need to support all operators in public hotspots.

Then there's backhaul. Putting an LTE femto on a 2Mbit/s ADSL line isn't going to be tenable, especially if it's from a third-party ISP which decides to throttle IPsec traffic at busy periods, as one example in Europe apparently does. You'd need high-end cable or VDSL or fibre to do justice to LTE. And outside Japan, I don't see much ubiquitous nationwide FTTH any time soon, given the economic outlook.

Last of all, there's the voice issue to consider. The current crop of early femto deployments from Sprint and Starhub tend to have heavily voice-centric homezone-type components to their business model. And yet, the question of LTE handset availability pivots on deciding what the LTE voice service looks like. Clearly, it needs to be some sort of VoIP - but what, exactly?

Overall, despite the attractions of the inside-out model, I'm not yet convinced it's truly the answer to LTE. I'm reminded of the words of a Nokia radio networks guy I talked to at 3GSM about 3 years ago, asking about femtos and picos. He said (in a gruff Finnish accent):

"Outside-in, always wins"