Indoor wireless coverage is coming to a cross-roads. In the future, will we use our own Wi-Fi networks, or will we have "femtocells" provided by the cellular operators?

At the moment, mobile operators provide voice connections outside the office and are persuading us all to upgrade to data in the wide area. But there's a problem: coverage. The 3G networks don't work so well indoors - and in countries like the US, even 2G voice can be pretty flakey.

The problem gets worse when data is used, and is worse the faster you want to use it. "In London, 3G coverage goes down by 66 percent when you go from 64 kbit/s to 384 kbit/s," , says Vicki Griffiths UMTS product manager at PicoChip, which makes silicon for femtocells.
For many, the answer is already there, in the fast indoor wireless networks we already use. Vast numbers of houses and offices have Wi-Fi networks, which have good coverage indoors and poor coverage outdoors. Fixed-mobile convergence schemes merge the two, so dual-mode phones can make calls over Wi-Fi indoors and 3G, GSM (or whatever other cellular network) outdoors.

There are two leading ways to offer Wi-Fi to cell convergence: unlicensed mobile access (UMA) and the SIP/IMS method. The two methods have been presented as rivals, with UMA favouring the mobile operators by routing calls onto their networks.

Why use femtocells?

But there's another idea: the femtocell, promoted by companies like Ubiquisys. This puts a tiny cellular base station inside the home or office (technically, indoor office cells are usually called "picocells"). Indoors, the phone roams to the femtocell, and makes calls as normal - but they are connected over the user's broadband service.

So which will it be, femtocell or Wi-Fi? Experts from important companies in the field got to grips with the question at a meeting in Cambridge, last week. The Future Wide Area Wireless SIG, organised by Cambridge Wireless and the Communications Research Network heard both sides of the question.

There are several purely technical reasons why femtocells might be a good idea. Any phone can roam to them because they use the same GSM or 3G networks. This is significant: one of the selling points of converged services is the promise of cheaper calls using the subscribers' own broadband networks - a saving which is currently wiped out by the high price of dual-mode phones.

But there's an even more powerful business reason why mobile operators want to sell femtocells: they hate Wi-Fi, because users own it and can use it at will. "Vodafone has no desire to subsidise a Wi-Fi handset," says Griffiths. Encouraging users to keep on using their mobiles indoors will keep them stuck in the grasp of their mobile provider. She believes that if operators price femto-based services right, customers will use cellular data indoors - and that will encourage them to carry on using it outdoors.

Femtocells should also slow down the operators worst problem: churn. Service providers can send users a femtocell when they threaten to dump the service because of poor 3G coverage in their homes.

If the operator manages to get everyone in a shared house or a family onto a femtocell, it's even harder for the whole family to move its mobile contracts - at the same time - to a different femto provider.

All this sounds very promising - but there are drawbacks to the femto model. The device is intended to prevent churn, and wean users off cheap VoIP services. It uses the customer's own broadband and phone line - so it's sitting alongside cheap or free alternatives. All this means operators will have to effectively give away femtocells. A large part of what the femto offers is based on bandwidth paid for by the user.

The femtocell is a key weapon for the operator in gaining access to the home, but the operator can't expect any revenue for data and calls in the home. It also has to be made very cheaply - for a cost less than $100. This is harder than might appear, because the radio link in cellphones is strongly asymmetric - the radio element in a femtocell can't be simply re-used from a handset.

Femtocells may make it harder for customers to change provider, but the cost of the femtocell has to be added in, so they increase the cost to the operator when a customer churns away to another service provider.

It's also not at all clear if the technology will be as simple as has been claimed. For example, customers will roam between macro cells outside the house and femtocells inside. But this means the radio coverage of the femtocell - managed by the customer - must be co-ordinated with that of the macrocell. And the operator can look forward to costly customer support calls, when the user unplugs the femto, breaks it or causes interference to it, - or when the femto fails because the customer's broadband goes down.

Wi-Fi Netheads versus cellular Bellheads

Like the old Internet-versus-phone company arguments, it boils down to Netheads versus Bellheads. "The Wi-Fi Netheads think 802.11 rules the world, but quality, security and battery life are tricky," said Mike Reynolds of technology consultant Sagentia, which hosted the FWAW meeting. "The 3GPP Bell heads have more handsets and better radio coverage."

A lot of people at the SIG plumped for Wi-Fi as the eventual winner. "There is a huge Wi-Fi ecosystem," said Trevor Coleman of CSR, a silicon maker that has a finger in the Wi-Fi and Bluetooth pies. As a free extra in broadband routers, and built into around 25 percent of phones by 2010, Wi-Fi is going to be everywhere, so it seems more efficient to use it.

The objections to Wi-Fi will go away: Wi-Fi chips currently add $15 to the price of a phone; in three years that will fall to $5. Power consumption and battery life will improve, as Wi-Fi silicon gets more sophisticated - only powering up parts of the chip when needed, for instance.
Despite this, the mobile industry has a habit of getting its own way. Even though mobile calls cost more, 30 percent of them are made indoors, and there is a steady shift from fixed to mobile calls - sometimes pegged at about 5 percent a year. Mainly the determination of mobile operators has led to strong predictions for their growth. Research company ABI predicts 102 million people will be using them by 2011.

Femtocells may look unlikely, but there's a possibility they may win out.