Debates about the future of mobile networks, and arguments about convergence between fixed and mobile devices (which are in fact the same argument) usually end up on the subject of IMS (IP multimedia subsystem), the next generation infrastructure of both mobile and wireline phone networks.

IMS, is ideally suited to fit the role of "technology panacea". It's widely agreed, and everyone can think of useful things to do with it. And, most importantly of all for the role of technology panacea - it doesn't actually exist yet.

Agreed by the 3GPP group which defines the future of mobile networks on behalf of operators, IMS supports IP applications on a telecoms network. It has spread beyond the mobile operators, becoming a key part of moves such as BT's 21st century network.

Converged systems run voice calls across both the cellular network and the Internet. Early services such as UMA are designed to operate without IMS, because it is not yet available on the networks where they run.

People pushing rival schemes have sometimes based their pitch on the idea that IMS combined with the session initiation protocol (SIP) is a better prospect than approaches like UMA, giving the user access to the full power of the Internet.

In fact, however, virtually all approaches to the future of telecoms networks are based on IMS. As UMA vendor Kineto explained to Techworld, UMA connections can support SIP and IMS applications.

Everybody's doing it
"IMS can be seen as a last ditch effort by the providers to prevent their becoming just a bit pipe," said Graham Finnie, senior analyst at Heavy Reading, at the NetEvents conference in Garmisch, Germany in February. As well as operators, other standards makers have adopted it, including the ITU in its Next Generation Network specification.

"IMS developed this momentum because it’s a very flexible architecture for delivering IP in any kind of access network," says Finnie. "It was developed for 3G but, because of the way it was designed, it was deliberately access agnostic, so in principle it can work in any kind of access network."

Finnie's research suggests that the majority of carriers intend to implement IMS between 2006 and 2008 - though he points out that intentions often slip. The main reason they want it, he says, is because it will allow carriers to deliver applications quicker, and get beyond the saturated markets for basic voice services.

But confusion reigns
Despite this, there is confusion. Only ten percent of operators believe IMS is clearly defined, 70 percent say work still needs to be done, and 20 percent believe it is still in the early stages of development. This is despite the fact that IMS is a published specification (this page sets it out a little more clearly than the raw 3GPP specifications page).

Thanks to the convergence bandwagon, IMS is often seen as a means to provide voice over IP, but this is not what it was originally designed for. "It was intended to be a mechanism to enable future services, but the focus initially was on non-real time services," says Dan Warren, senior core network architect at Vodafone, and a spokesman for the Multi-Service Forum, a group of operators and equipment vendors working on practical implementation guides to complex telecoms standards like IMS.

"Concerns like lack of vendor interoperability, vendor lock-in, quality of service agreements, concerns about security and reliability will re-emerge," says Finnie. "The radiant glow in which IMS is currently sitting, is likely to be dissipated somewhat by the experience of actually trying to deploy the technology."

IMS doesn't provide end-to-end quality of service, says Warren. "That's not due to IMS itself, but to the enablers in the network - the way QoS is pushed down into the access part of the network. It is particularly complex where the access network isn’t owned by the same operator or carrier as the subscriber is using for their IMS."

"If you are making a call from one IMS domain to another IMS domain, the interface between those IMS domains is very technically complex, because there is a huge amount of flexibility in VOIP services. It is also very commercially challenging as it becomes a voice over IP inter-connect." In other words, IMS doesn't escape from the reality that multiple network providers carrying a call will want to charge each other for a level of quality.

Another problem is allowing the signalling path - that sets up the call - to separate from the bearer path - which carries the data. A call between two people in Germany might require signals to take a very long route to get to the users' home networks in the UK and the US. Currently, the call data would also have to follow that route, because that's the only way quality of service can be applied, says Warren.

How will it play out?
The MSF will be running an interoperability test in October which should iron out some of the technical issues in getting IMS systems to work together. At the same time, the commercial issues will come into focus.

"Older networks are coming to the end of their life, and operators want to reduce operating costs within their networks," says Mac Taylor of analyst firm Moriana Group, which has produced a very large free report on IMS. "They want to create new services quickly and drive new revenue. But there are other ways of doing that besides using IMS." The same sorts of things were promised three years ago by the Parlay proposals, and fifteen years ago by intelligent networks (IN).

"IMS is not simply a network upgrade from [the current signalling system] SS7," says Taylor. "It’s not an SS8. It’s actually a fundamentally different way of looking at the Internet. It’s trying to achieve what the internet has achieved, but risks actually destroying some of the freedom and innovation of the internet." By not being a simple end-to-end platform, IMS could stifle the innovation that flourishes on the Internet.

"If Adobe had invented PDF in the days of IMS," says analyst Dean Bubley of Disruptive Analysis, "we'd be paying every time we downloaded a PDF file."