For those who can remember it, there was a time when cellular backhaul was boring. Not these days. Backhaul has been one of the hottest segments in telecom for three or four years now, and so it remains. One segment of the evolving cellular backhaul technology story that is only now about to go through the same transformation cycle is network synchronization.

While some CDMA operators use GPS for synchronization, most cellular networks today derive their synchronization from an E1 or DS1 at the cell site. But deploy packet backhaul at the cell site – as most operators are doing to support the waves of 3G data that are hitting their base stations – and a whole different modus operandi presents itself. Or more accurately: a whole range of different modi operandi.

Remove the DS1 or E1 from the cell site altogether without introducing a new synchronization solution, and the mobile operators' voice service will deteriorate – imperceptibly to start with, perceptibly over several weeks or months. That's why the majority of cellular operators still prefer to leave DS1s or E1s at the cell site for voice services and synchronization when they introduce packet backhaul: That way, they get some of the cost savings associated with packet transport for backhauling their data traffic, while leaving their voice network undisturbed.

The issue with that approach is the "some" in "some of the cost savings." This is because leaving TDM at the cell site when introducing packet backhaul for 3G data traffic can reduce the scope for cost savings by as much as a third compared with an all-packet backhaul environment.

The two standards that have been developed to address synchronization in this environment are Institute of Electrical and Electronics Engineers Inc. (IEEE) 1588v2 and the International Telecommunication Union (ITU) G.8261 Synchronous Ethernet standard.

On paper, most operators that Heavy Reading has worked with – particularly those in mature markets – clearly prefer Synchronous Ethernet. This is because it conforms to the L1-based approach to synchronization that they are used to with TDM. The big challenge with Synchronous Ethernet, however, is that it needs to be supported throughout the network, requiring upgrades to a large number of network elements. So while some big incumbents, such as Deutsche Telekom AG (NYSE: DT) and BT Group plc (NYSE: BT; London: BTA) are moving ahead with deployments of Synchronous Ethernet in the coming year, the standard is still some way off from scaling in large commercial volumes from a global perspective.

The IEEE's 1588v2 standard is a Layer 2-based solution. There are several working implementations of it in live commercial service already today, such as by Vodafone Portugal and Telus Corp. (NYSE: TU; Toronto: T) (Canada). But again, there are challenges with 1588v2. The first is that as a Layer 2-based solution, 1588 requires a first-class Layer 2 network underlying it to perform optimally. A Ferrari driven over a rugged terrain is still a Ferrari – but through no fault of its own, it won't perform as it's supposed to. Moreover, lack of standards in the implementation of the 1588 algorithm itself means that while intensive bilateral interoperability efforts between two vendors can result in a workable solution, scaling that up easily into a mass market of networking products whose 1588 implementations are all interoperable is taking longer than many operators would like.

Two years ago, a lot of operators were giving Ericsson AB (Nasdaq: ERIC) a hard time for advancing its own proprietary synchronization solution that leverages NTPv3. At group level, for example, Vodafone Group plc (NYSE: VOD) wouldn't hear of it. Now Ericsson has Telstra Corp. Ltd. (ASX: TLS; NZK: TLS) (Australia) and KPN Telecom NV (NYSE: KPN) (Netherlands) among a dozen or so operators that are deploying it.

Operator and vendor complaints that proprietary approaches could prove difficult to operationalize and will create vendor lock-in may contain a large grain of truth in many cases. But in the context of the challenges that are also presented by other standards-based alternatives, these criticisms don't carry quite the same weight as they did two years ago.

Finally into the mix we need to throw the Long Term Evolution (LTE) requirements to support phase as well as frequency and time-of-day synchronization. This may only be required for Network MIMO and MBMS-SFN in future 3rd Generation Partnership Project (3GPP) releases, so it is debatable exactly when this will be required, or even whether some cellular operators will need it at all. Then again, any operator looking to support phase synchronization will find that the only candidate technology with a proven track record in supporting it is GPS. Phase synchronization is something that Synchronous Ethernet is never likely to support, and the IEEE has barely started to look at how it might be supported in 1588.

This support for phase synchronization as well as frequency and time, combined with the slow rate at which other standards-based packet synchronization solutions are scaling, explains why GSM and W-CDMA operators throughout the world are increasingly open – or to put it another way, less opposed – to leveraging GPS or other future Global Navigation Satellite Systems (GNSS) for their packet backhaul synchronization.

The challenges that operators face in selecting a synchronization strategy for packet backhaul – and then scaling it – will be the subject of one of the panel sessions at Light Reading's upcoming virtual tradeshow on February 4, Packet Backhaul 2010: Scaling Up to Bring Costs Down. Hope to "see" you there!

— Patrick Donegan, Senior Analyst, Wireless, Heavy Reading