Return of Low Power GSM

About two years ago, there was a sudden buzz of excitement around a new spectrum auction in the UK - referred to as "low power GSM", it represented a sliver of frequencies that were formerly a "guard band" between GSM @ 1800MHz and DECT. There wasn't enough spectrum to build out a national network, and it was expected that instead it would just be used for low-power indoor coverage, typically with picocells. And rather than selling the spectrum to a single operator, it was suggested that multiple indoor providers could share it - with operators expected to work with each other to avoid interference.

Ofcom sold 11 LPGSM licences for quite low prices, to an eclectic range of companies, from BT (BT) and C&W though to Swedish operator Spring and some very random ones like Philippines Long Distance Telecom.

Much excitement ensued, with speculation (by me among other people) that this was another interesting FMC route, competing with WiFi-based alternatives, using ordinary GSM phones. I'd been following the use cases for picocells since 2001, and this was quite a milestone. (Bear in mind that May 2006 was before the femtocell hype started to hit home). But.....

Not much has happened since. The big problem is that having private indoor cellular networks is a great concept, but runs into practical and user-experience problems when people leave the building. There are four options for what happens next:

• The phone stops working. The SIM card is for the indoor-only operator.
• The indoor service provider also has a macro service network outside, and the phone just hands over.
• The user does a manual "network reselect" on their handset, either using two SIM cards or with the indoor operator doing something clever with a small offshore roaming service provider.
• The indoor operator signs a "national roaming" deal (ie becomes an MVNO) with one of the main national MNOs for outdoor coverage.

The only UK LPGSM operator with its own macro GSM network is O2. But it seems to have been singularly disinterested in doing anything with its license, and perhaps just bought it cheaply to check up on what everyone else was doing.

Two licensees already had MVNO deals in place that could have possibly enabled them to knit together overall solutions. But BT has been more focused on Fusion and dual-mode WiFi-cellular, and Carphone Warehouse probably doesn't have the enterprise reach needed to justify the price of picocells. [Sidenote: picocells cost around $2k apiece. And most LPGSM opportunities are on corporate sites that would need integration with existing PBX and LAN infrastructure.]

Until recently, Teleware's arm Private Mobile Networks was the only commercial LPGSM operator with a launched service and growing channel. I met with the company recently, and it has also developed a productised version of its solution tailored for the construction industry. Teleware's background is in developing communications servers & applications for the enterprise market, so it has got a lot of the important systems integration and channel-partner credentials. (And also, in the construction industry, building sites obviously don't have existing PBX/LANs).

My understanding is that MVNO terms from the major operators have been less-than-generous. Understandable, given that the LPGSM model represents a possible cannibalisation threat for their corporate business. Also, the timing of roll out of corporate dual-mode WiFi FMC was also delayed, so this more cellular-centric approach hasn't even been considered as useful hedge. That situation has changed recently however, with real-world dual-mode deployments in some UK companies (eg BT Corporate Fusion) & education facilities.

But now, C&W has finally cut a roaming deal with Orange to become another live LPGSM operator. (see the press release and also articles like ZDnet's). It's intending to launch a corporate FMC solution by the end of 2008.

One of the interesting things is that C&W's recent business update presentation refers to its supplier for the solution as Ericsson (ERIC). The significance is that Ericsson has long talked up its 2G femtocell (rather than picocell) products. Most of the femto vendors have been focused on 3G rather than 2G because of a different set of challenges in network integration, and the fact that 2G coverage is generally pretty good anyway. Presumably, this will bring the capex requirement per-site down from a few thousand pounds towards a much lower figure.

I've also heard rumblings that a few of the other LPGSM providers are also stirring from their slumber. Having 2G femtos available as well as picos should definitely help. I suspect we'll see a few more announcements over the next few months.