September 2004

Migrating from TDM to IP
Real-World Business and Technology - The Mobile Persepctive

BY BROUGH TURNER

In the past ten years, mobile telephony has been really hot, while IP telephony has been mostly hype. Now IP telephony appears to be entering widespread deployment — IP-enabled PBXs are outselling traditional PBXs and residential VoIP services are being discussed in the popular press. But VoIP and mobile telephony still don’t mix. Why not? How long will it take? And why should readers of INTERNET TELEPHONY® care?

Mobile telephony is exciting because it’s growing — rapidly! In the past seven years, mobile telephony has gained more subscribers, worldwide, than fixed telephony acquired in 120 years, with the result that today mobile subscribers far outnumber fixed subscribers (~1B to ~1.4B worldwide). And growth continues at a fast pace. Mobile telephony is also exciting because it better addresses our human need to communicate — you call the person you want to talk with, not the place you think they might be. Finally, like PCs, mobile telephony benefits from Moore’s Law improvements in integrated circuits. And it benefits from similar, ongoing, exponential improvement wireless efficiency, i.e., in bps per MHz per sq. km.

But Moore’s Law applies to only the basic technology. While it continuously impacts established markets — PCs have more memory and run a bit faster every year, for example — changes in major infrastructure — TDM to IP, for example — take much longer as they require major shifts in investment. Likewise, the adoption of new applications, or even new features for traditional applications, can take years if they require shifts in human behavior.

So while converged networks are happening on the wireline side, mobile network convergence will take longer, especially at the radio link or “air interface.” It’s relatively easy to deploy VoIP over a broadband connection. It’s a lot more difficult when bandwidth is precious as it is for mobile operators.

The GSM community — with 70 percent of the world’s mobile subscribers — has defined a migration path that will eventually get them to VoIP. The GSM version of VoIP is called IMS (IP multimedia subsystem) — part of the Third Generation Partnership Project (3GPP)’s Release 5 architecture. But that’s two releases ahead of what’s being widely deployed today, i.e., Release 5 deployment is still years away. In any event, services are more important than whether voice uses IP or TDM infrastructure.

Services Versus Infrastructure
Most countries have multiple mobile operators, so markets are either competitive, or fiercely competitive. This means call quality is important, driving infrastructure investments to provide coverage and capacity. And business efficiency matters, driving cost savings wherever feasible.

But the primary competitive differentiator — and the only source of new revenue — is new services. Today, the newer revenue sources are ringtones, text messaging (SMS) and data and multimedia applications that leverage IP technology. In the near future operators expect to gain revenue from ringback tones, multimedia messaging (MMS) and video.

New Services Require a Mix of New and Old Infrastructure
SMS sends text messages over the traditional SS7 signaling channel and ringtones are typically downloaded to the phone via SMS. MMS uses IP transport, but ringback tones are a network service based (at least today) on traditional Intelligent Network (IN) technology. Internet access is via IP, of course. And the newest video services use a mix of messages over MMS, streaming over IP and live video telephony over circuit-switched data! Yes, you read that correctly. The new 3G mobile video telephony service that is taking off in Japan uses 64 kbps circuit-switched data, not IP packets.

Someday all of these services will run on a pure IP network, but parts of that solution are still being invented. Meanwhile, pressure for new services is here now. Luckily, we can apply design principles from the pure IP world to help introduce new services into mobile telephony’s mixed network infrastructure.

Decomposed Network Views
Before the advent of IP telephony, traditional telephony was moving to IN where application processors (Signal Control Processors — SCPs) and media servers (Intelligent Peripherals) support new services by interacting with the traditional switching systems.

With the advent of VoIP, two competing models for IP-based services emerged — the softswitch model and the end-to-end model. Variants of the softswitch model appear in the work of the IPCC, the ETSI TIPHON project and the 3GPP. The most vocal supporters of end-to-end VoIP are the SIP community within the IETF. In the softswitch model, enhanced services are provided using application servers and media servers. In the end-to-end model, a media server is just another endpoint, but one that has capacities and features well beyond the typical SIP telephone set.

All approaches divide the work of providing a telephony service, assigning specific functions to specific subsystems. While the names on the blocks are different, the functions are remarkably similar across the different network views. And, at least for the two VoIP approaches, the subsystems are interconnected using IP networks and open, IP-based protocols.

For new mobile deployments, we’d like an IP-based service delivery architecture that can support all network variants at once — TDM/IN, softswitch, and SIP. This is not so far-fetched. The primary differences are in signaling and in media connectivity. If we abstract the signaling functions (and some media adaptation) into a gateway services component, the rest of the application services and media services can be the same for TDM, softswitch or SIP networks. A common platform is feasible, as I’ll demonstrate in the context of a specific service.

Page 1 of 2 [Go to Page 2]