About 18 month ago, I wrote a deliberately provocative post titled "The Tyranny of the SIM Card". It tackled the thorny issue of whether the SIM (which is essentially 'an operator on a chip') should maintain its central role as old-school cellular telephony moved towards FMC or triple/quadplay.

In an FMC world, if you don't have a SIM in everything that you connect to the network (phones, PC, TV etc), does it still make sense to have a SIM in anything?

A discussion I had recently with a company called Blueslice (which primarily makes advanced variants of HLRs for mobile networks) has sparked off a revision of my thoughts.

I think the SIM's usefulness is gated by its name - an Identity Module. This is valuable for when I need to have a single, consistent identity associated with a service. When I want the service provider's infrastructure to be able to recognise me "Ah, that's Dean Bubley trying to connect".

The most important use case I can think of in mobile comms, of having a single, consistent identity is having a consistent phone number.

But equally, there are also plenty of instances where the user might want multiple, or perhaps temporary identities to be associated with a single device. A classic case is mobile broadband access from a PC. Yes, I might want to have an ongoing account relationship with a given HSPA service provider - but I may also want to get temporary access to a second, third or tenth provider because of the specifics of local availability, price, speed and so on. I'm not taking inbound calls, so I don't need consistency of identity - I just want local connectivity. (And perhaps local services & content too - this isn't a 'dumb pipe' rant).

Yes, roaming performs the function too - but it's expensive & hugely complex. And unnecessary much of the time. If it weren't for the identity aspect of the phone number, why would I want my home operator to "lend" my identity to another service provider? If I use an Internet Cafe, they don't bill me via my home broadband. If I rent a car in the US, it's not added to my UK car lease. When I stay in a hotel room in Spain, they don't contact my London landlord to add the cost to next month's rent with a 10x markup.

Sure, there are exceptions when certain aspects of security and log-on convenience are involved, and so companies like iPass can make sense for business users. And yes, you can always buy a local operator's SIM card and put it in an unlocked device, although that's not always convenient either, especially for data access.

And even more counterintuitive is the notion that you would use a clunky mobile SIM card to authenticate for services or devices that are usually SIM-free. A classically pointless example has been the use of EAP-SIM for registering for WiFi on a laptop. Well, that's fine if I actually want to use a mobile operator's own WiFi hotspots, but what about the other 90% of the time? It's just another attempt to use a hardware form of lock-in on what has traditionally been an unlocked device. I've criticised SIM-based WiFi authentication for years, and my opinions seem to have been vindicated as it's shown very limited traction to date.

The thing I liked about some of Blueslice's usage cases were that they allow operators to be much more pragmatic. For example, combining phone+SIM authentication for voice, plus laptop+username/password authentication for WiFi - in the same subscriber record in the HLR, but with 2 separate profiles. This makes a huge amount of sense, especially as we move towards multi-access infrastructures. It seems that most WiMAX devices won't be SIM-based, so combining a single user account spanning GSM/UMTS plus WiMAX services might otherwise be tricky.

Other solutions to this class of problem include a growing number of dual-SIM phones, or "soft SIMs" of which a given device might support several simultaneously.

Bottom line - don't force physical-card SIM authentication onto products or services where the user might reasonably want to have multiple or temporary identities, or where roaming is an overengineered solution. Think about the growing likelihood of customers having multiple devices, some of which might be operator-specific (eg a subsidised & customised phone) and needing consistent identity, but also others which have a much more ephemeral connection to a given operator and for which using or obtaining a SIM is an unnecessary burden.

[Edit - apologies for overuse of the word 'tyranny' in two consecutive posts - writing the first item made me remember to do an update on the second]

Original Source: Dean Bubley's Disruptive Wireless