Why WLans and 3G must settle their differences

Two technologies can chase one market if roaming is enabled.

Two technologies can chase one market if roaming is enabled.

The relationship between public wireless Lans based on 802.11x and next-generation mobile networks is a complex one.

WLans pose a clear threat to the mobile market. The mobile industry has largely bet its future on data, believing the ongoing decline in the revenue for voice services can be offset by selling customers new services based on data and multimedia. It has invested heavily in developing, and acquiring spectrum for, new technologies that offer higher data speeds and the capacity to support them for many customers.

Then, out of the blue, comes a new technology that is apparently cheap, uses unlicensed (and therefore free) spectrum, and offers data speeds that make next-generation offerings look rather pitiful.

But there is also synergy. WLan provides high data rates but in limited areas. Next-generation cellular provides lower rates over far wider areas. It's easy to see how end-users would benefit from a device and a service that could be used in both contexts - with WLan where available, but falling back on the cellular network's data service where it was not.

A multi-technology device is needed; something that can function on both kinds of network. Nokia has for some time offered dual-mode PC cards that can do both GPRS and WLan, and contain a Sim card reader so the cellular network can recognise and bill the user. And other suppliers boast similar products, either on sale or in development.

However, the real key is not devices but roaming - the ability of customers of one network to use the services of another, typically through prior inter-network agreements so that the customers themselves don't have to do anything special.

Since the advent of GSM, cellular customers have been rather spoiled because this capability is designed into the very foundations of the system. But for the emerging public WLan industry, this is all still TBD. Indeed, the industry uses the term roaming in at least three different ways:

n Intra-network (or access point) roaming - being able to move from one access point to another in the same "hotspot" deployment without losing the session (what cellular people normally call hand-over)

n Inter-network roaming - being able to use the services and coverage of a public WLan even though you are registered as a customer of a different one

n Cross-network roaming - being able to use the services of a public WLan even though you are registered as a subscriber of a cellular network (or vice versa).

This is compounded by the lack of a definitive business model for public WLan. Cellular was designed as a public service to be offered to a paying public, with authentication, usage recording, charging and billing principles built in. WLan was not - it is a link layer specification, using wireless as a cable replacement. So it doesn't have any of the necessary commercial infrastructure as part of the standard. You can add it in, and you can do it in any number of ways.

Sure, some operators are clear that they intend to sign up subscribers, authenticate them, record their usage and bill them. But others may only provide service to customers as they turn up, through credit cards or scratchcards, say. Still others talk of offering service for free, to attract customers into their establishments, or as part of an overall premium service package.

This makes it hard to arrange roaming agreements between local public WLan providers - even though the ease of setting up a local service means fragmented coverage is inevitable. Nor is there much agreement on whether roaming should be through many-to-many bilateral arrangements, a multilateral hub-and-spoke arrangement, or some sort of aggregator role.

So it's not easy to see a way through for commercial models needed for public WLan-cellular roaming. Cellular operators need to authenticate "visiting" users when they turn up on their network and know where to send the bill. The Sim card and the numbers stored on it are what is used for this. Unless WLan operators use a similar framework, agreeing principles for roaming is going to be a hard slog.

One way out is for the public WLan industry to adopt wholesale the cellular service management and commercial infrastructure - authentication protocols et al. Some suppliers, notably Nokia, are pushing for this, and some operators are adopting it - even though it is a proprietary technology at the moment. There is also work on a standards-based approach, catchily titled Enhanced Authentication Protocol-Sim.

This goes against the ethos of the public WLan industry, with its unlicensed guerrilla credentials and its ability to make de facto standards on the fly - unlike the telecoms "dinosaurs". But in the longer term, the big boys of telecoms will move into the WLan space. If it is to be successfully monetised, it will become a more orderly, boring place. But we might get roaming that works.

Jeremy Green is research director for Ovum's Wireless Analyst Group

Read more on Wireless networking