I've been at the Marcus Evans UMA conference in London the last couple of days. It's been a bit of a requiem, to be honest - unsurprisingly, I put my boot into the unloved technology pretty hard once again, and I was joined by a gang of other naysayers (notably the always-entertaining Alan Duric of Telio).
There was a smaller-than-usual crew of scorned UMA-acolytes pretending that all was fine - Kineto (obviously), plus a few other vendors. Nokia's representative did his best to give the company's usual grudging endorsement of the technology, essentially "It's OK for some operators, in some contexts", with the usual undertone of "...but we'd much rather you bought lots of IMS gear from us instead". Ericsson and Alcatel were pretty under-represented at the event, given their enthusiasm, while there was a conspicuous lack of lots of operators eagerly scribbling down the details of potential new dualmode FMC services.
There were some noteworthy things to come out of it, though. I've been saying for a while that UMA may get reincarnated as a more generic authentication technology, exploiting its stronger points around security, while avoiding the huge difficulties in creating phones. It was very notable that Kineto & others referred to embedding UMA agents within non-phone devices like terminal adapters and cellular pico/femtocells (tying in with a discussion I'd had with femtocell advocate Ubiquisys earlier this week). The big advantage of this is that these devices don't have user interfaces & lots of applications that need to be made "dual-mode" friendly. There was also an enterprise-centric pitch from NET, which I need to dig into a bit more, but which at first sight appears to only have a chance in a niche of a niche of a niche. Of course, none of these non-phone UMA widgets actually move in & out of cellular coverage, which underlies the irrelevance of the much-vaunted cellular-WiFi handoff bit of the technology.
On that theme, I've been leading a campaign against the term "Seamless" for about 3 years, insisting that "Seams are important" in networking, and that it's stupid to try & get rid of them.
So I was deeply impressed by a BT presentation, which as well as talking about Fusion, touched on IEEE 802.21 Media Independent Handover, and talked about "Intelligent Handover", wherein the applications, network & device take account of the properties of the different networks available to them and change their behaviour accordingly. So, instead of the application being "bearer agnostic", it is "bearer aware", and for example changes codecs, or alters the way that video is being transmitted, based on available bandwidth/latency/cost and so on. In other words, it starts to fix the biggest underlying flaw in both UMA and IMS philosophies.
The BT representative has also shown the operator has had another epiphany, and is trying out different types of mobility management in the lab - switching between WLAN, cellular, WiMAX etc based on either centralised control in the network (ie as UMA works now) or uses the device itself to monitor local network conditions (signal strength, congestion etc) and make decisions autonomously about how/which to connect to. At present, the demo setup uses a laptop rather than a phone to do this, but the solution should scale down to smartphones over time.
It strikes me as deeply ironic that it has taken a fixed operator to recognise the benefits of Moore's Law on the mobile device, and that instead of being a "dumb terminal", a handset actually possesses the ability to make network connectivity decisions based on its own local conditions, and the applications being used on it at the point in time.
The cellular community, on the other hand, remains ruled by the radio network & switching departments, which largely refuse to recognise that software and chips on the actual phone could have any useful input to controlling or defining a service. Which is why we have had a continued stream of network-derived innovations, which have fallen down when people have actually tried to implement them on phones & persuade people to use them (MMS, videotelephony, and IMS in the future). UMA itself is a very clever radio network solution, aimed at solving problems in the access network, but created without regard for the ultra-painful "collateral damage" involved in actually getting the user experience right.
It is also, no doubt, why the visionary 802.21 concept is coming from the IP community, rather than 3GPP, which remains mired in the wasteland of bearer-ignorant applications.