Backhaul Timing: Anything But Synchronized

Yes, absolutely. Maybe I should have spelt out the "Any 2 Will Do" trend I mentioned at the foot of my last message in more detail. Increasingly many operators see that no one solution will provide a full-proof solution for all their current and future cell site synchronization requirements. Hence increasingly operators are looking at deploying two in tandem - such as the Synch E and 1588v2 combo that you allude to.

As for Synchronous Ethernet and 1588v2, why not both? It seems to me that it is quite a lot simpler and less error prone to keep the frequency synchronized using Synchronous Ethernet. Just about all you have to do is lock to the incoming Ethernet clock, whereas with 1588v2, there is all kinds of message passing, averaging and filtering, and many ways for it to go wrong.

And once you are synchronizing frequency with Synchronous Ethernet, then 1588v2 becomes much simpler as a method for synchronizing time.

Any idea as to whether operators might go in this direction?

Written in January 2010, my column above was clear at that time that operators "clearly prefer Synchronous Ethernet" among standards-based synchronization solutions for Ethernet backhaul.  And I had very recent research evidence at the time time to back that up.

Over the last 18 months, however, market sentiment has shifted. There is now at least as much, if not more, momentum behind the IEEE 1588v2 standard as there is behind Synchronous Ethernet. 

One key factor in bringing about this shift has been the ITU's commitment to develop a roadmap for phase/ time as well as frequency synchronization for IEEE 1588v2 whereas there is no such roadmap for Synchronous Ethernet which will continue to support only frequency synchronization.

The level of industry energy going onto 1588v2 for Ethernet backhaul is increasing markedly now - for example via Symmetricom's SyncWorld partnership program as well as the IEEE's 1588 Conformity Alliance which is about to begin conformance testing for 1588v2. The Conformity Alliance also has plans to begin serving as a conformance test house for the Synchronous Ethernet standard as well, providing a potentially valuable one stop shop for equipment vendors.

Again I have recent new research evidence to support this recent shift in sentiment towards 1588v2 and it will be used in both my upcoming White Paper and Webinars relating to Light Reading's new 1588v2 Industry Initiative. When it comes to choices of synchronizaton techniques, "Any 2 Will Do" is becoming a new industry mantra. This highlights that whilst each of the various new synchronization approaches have their own strengths, on their own they can each have their limitations as well.

OCNW offering solutions for this space as well.

http://event.on24.com/event/18/92/80/rt/1/documents/slidepdf/01282010_2010outlook.pdf

See slides 14, 15, 24

The newer base stations can be said to be Internet Protocol (IP) rather than TDM based; the L2 protocols (MPLS/Eth/PPP) below IP do not really make that much difference compared to the shift to end-to-end IP networking. But for the synchronization delivery, which L1 protocol is used makes a difference; hence my preference for IP over L2 MPLS, over POS as long as the fiber goes.

I don’t see the reason to move from the existing synchronous L1 protocols to new ones, especially ones to which the synchronization features would need to be retrofitted.

Newer basestations are Ethernet based (and most are because T1/T3 is too slow) and will increasingly have Synchronous Ethernet and 1588 built in, since it costs very little to do so. I think it is your POS idea that would require a lot of retrofitting of emerging Ethernet based backhaul and basestation equipment.

My understanding is that at present, basestations that are installed in locations with only an Ethernet connection are using GPS for synchronization and time of day. As Synchronous Ethernet and 1588 are more fully developed, they can save the cost and reliability concerns of using GPS and have cheaper basestations.

Yes, the availability of the (sub-physical-layer, L0) pipes supporting 100Mb/s and higher bandwidths to the cell sites is the #1 cost problem: the cost of installing fiber will largely dominate the upfront cost of getting the required (3/4G) bandwidth connectivity to the cell sites. And the cost of running fiber to the cell sites is the same regardless of the L1/2 protocols used; pure-packet protocols like Ethernet do not bring any meaningful savings here.

But once there is the business case (read: necessity) to run the fiber to cell sites, I don’t see the reason to move from the existing synchronous L1 protocols to new ones, especially ones to which the synchronization features would need to be retrofitted.

One can carry IP/MPLS packet traffic just the same over POS as over optical Ethernet, and in current hardware design, the physical hardware (bit transparent optical transponders combined with a large FPGA) will be just about the same regardless of the digital protocol layers used.

So I would tend to prefer existing synchronous carrier signals as long as the fiber goes (or there is biz case to run fiber), and microwave and/or plain old PDH copper (T3) from there on.

Reasonings for alternative approaches welcomed too.

Patrick,

Your article implies that there are only 12 networks and they are only done by the big E.

AT&T has a multi-vendor environment with Cisco as the other vendor (and they as well have a method to make this work).  So, I think that there is a lot of skepticism around 1588 but Ethernet backhaul proceeds apace.  The biggest issue is still availability of Ethernet physical pipes.  

Since Tellabs has already announced revenue from BT, you probably need to take another look there - like perhaps Ethernet Radios for backhaul.

seven