Where's PON Going Next?
Yet where PON really goes from here is boggling the minds of both standards-setters and business strategists: In research for a new white paper on NG PONs, we found that the road ahead is by no means as clear as the 10G standards might suggest.
On the face of it, PON evolution looks an easy equation: Work out how much bandwidth you think you will need five or ten years from now, and buy the technologies that fit the projection. On a simple reading (i.e., just projecting out the likely residential downstream connection speed), operators will need to offer 200 Mbit/s to high-end customers by 2015, and 2 Gbit/s by 2020.
GPON can't offer 200 Mbit/s uncontended to every customer, and 10GPON can't offer 2 Gbit/s uncontended. Solution: NG PON 2 -- on which the Full Service Access Network (FSAN) group has just begun work -- sometime between 2015 and 2020. In the next five years, NG PON 1 (a.k.a. XGPON or 10GPON), will be just fine for those that need an upgrade from GPON. (On the EPON side, 10GEPON will fulfill the same mandate.)
But this analysis conceals a whole host of issues.
First, any calculation will be upset if -- as some hope -- mobile backhaul and business traffic starts filling up PONs. That certainly hasn't happened yet, but operators naturally would like to offset the massive cost of fiber rollout by connecting new classes of users. If that happens, it will both increase overall requirements, potentially leading to an earlier migration to NG PON; and, more importantly, require a flexible technology that can meet both residential and business bandwidth requirements and the needs of businesses for symmetry, security, redundancy and so on.
Secondly, the fact that PON bandwidth is shared means it is fundamentally different from DSL. Calculating on the basis of uncontended bandwidth per user thus underestimates what it is really offering. In reality, the 2.5Gbit/s downstream and 1.25Gbit/s upstream bandwidth may be enough to carry operators well into the second half of this decade, so long as everyone doesn't want everything at once.
There are other considerations, as well: For instance, some operators are pondering a massive consolidation of central offices, and they are looking for longer-reach technologies that go as far as 100km. This completely alters the business case -- and therefore the way they look at the options today.
On top of all that, the current legacy base varies enormously from one country to another. Verizon Communications Inc. (NYSE: VZ) has passed 16 million homes with PON; BT Group plc (NYSE: BT; London: BTA) has barely begun. The gap in years between operators continues to stretch. And then there's the global technology fragmentation issue, with major Asian territories still deploying to the Institute of Electrical and Electronics Engineers Inc. (IEEE) GEPON standard, and many European fiber builders opting for Ethernet P2P, the technology that also still dominates in the fiber-fed business market.
No wonder some operators are considering sitting out the 10G stage and waiting for that next standard. The obvious payoff: less disruption, and fewer technology generations.
If 10GPON does turn out to be a niche technology with little purchase in the market, it will increase pressure on standards bodies to agree on a new standard. FSAN, which creates proposals for the International Telecommunication Union (ITU) to formally standardize, has already begun work on NG PON 2, but there is a long way to go. Many ideas have been proposed, and as usual there are plenty of vested interests with their own pet ideas to push.
All the same, some of the smoke around NG PON 2 is already beginning to clear. A tentative timeline has been set that envisages publication of a standard around 2013. And some of the many ideas put forward, such as 40G TDM PONs, are likely to be discarded. In fact, it seems increasingly likely that NG PON 2 will be based to a greater or lesser degree on WDM.
Unfortunately, there's a caveat here, as well: There are many proposals around WDM, and they tend to entail different evolution strategies. For instance, one that has strong support from establishment vendors as well as some incumbents is stacked or hybrid PONs -- in which the 10G elements are retained, but "stacked" into four (perhaps more) wavelengths. Another proposal is for a far more ambitious scheme that uses coherent transceivers to deliver as many as 1,000 wavelengths, all carrying 1 Gbit/s. And there are plenty of other ideas in between.
The considerable effort now going into resolving these issues leaves us optimistic that the next step will be reasonably clear by 2013. In the meantime, with even the most enthusiastic 10G supporter, China Telecom Corp. Ltd. (NYSE: CHA), predicting no significant deployment before 2013, expect the "legacy" PON technologies to have a good run yet.
— Graham Finnie, Chief Analyst, Heavy Reading