by Stephen Hardy
While speakers at Lightwave's Optical Access '08 e–conference see a bright future for 10G PON, most see major hurdles to near–term acceptance of WDM–PON, mainly because of cost and deployment challenges.
The talk at Optical Access '08, Lightwave's e–conference held December 4, focused squarely on the next generation of fiber to the home (FTTH) and building (FTTB) infrastructure. Not surprisingly, speakers discussed 10–Gbps PON enthusiastically. The tone regarding WDM–PON, however, was decidedly less upbeat.
Vincent O'Byrne, director of FTTP architecture and design at Verizon (www.verizon.com), established the appeal of next–generation (NG) PON in his opening keynote address. O'Byrne asserted that his company's current GPON architecture, which reached general deployment only at the beginning of last year, should meet Verizon's needs for the foreseeable future. That includes the roll out of 100–Mbps services; while O'Byrne gave some (including us) the impression that the carrier would introduce such a service this year, O'Byrne and Verizon Telecom Director, Media Relations Jim Smith subsequently told Lightwave that the carrier would merely make the engineering choices and construction in 2009 to enable 100–Mbps services. “It is not on anyone's actual product rollout plans at this point,” Smith wrote in an email in December.
Regardless, the next generation of PON technology, specifically the work now underway by the Full Service Access Network (FSAN) group and the ITU–T on 10–Gbps GPON, is on O'Byrne's mind. “We are looking at what's coming next, how we should possibly transition, [and address] some of our concerns with the technologies being proposed,” he stated.
O'Byrne noted that FSAN and ITU–T are looking at two flavors of NG PON. NGPON1, which is on the front burner, focuses on the 10–Gbps efforts; NGPON2, which some have called “future PON” work, covers such technologies as WDM–PON. NGPON1 itself has two variants. XGPON1 specifications cover 10 Gbps downstream and 2.5 Gbps upstream. Symmetrical 10–Gbps transmission falls under XGPON2. (For Verizon's perspectives on NG PON, see the table below.)
Of the two, O'Byrne says XGPON1 currently holds the most appeal. However, it will take more than the ability to increase bandwidth for Verizon to deploy NG PON, O'Byrne revealed. “One of the items that is important for us, given that we have deployed GPON already, is the requirement that it be able to coexist with GPON on the same fiber,” he said. “In a lot of our deployment cases, if we can support it via some of our IT systems, it will actually reduce costs if we can do it on the same fiber, especially at low take rates.” Given the cost of deploying a new PON—because of the use of an RF overlay scheme to provide broadcast video services, “every time we add a new PON, we have to light up all 32 splits with an EDFA, and that increases the cost considerably”—O'Byrne said that in brownfield applications, Verizon would most likely deploy a standalone NG PON only where GPON bandwidth has approached exhaustion.
Cost, of course, is also a primary concern for Verizon. O'Byrne said that input from the vendor community leads him to believe that it will take about five years from initial deployment for NG PON costs to attain parity with current GPON. Commonality between NG GPON specifications and those of the IEEE's 10G EPON would provide a catalyst toward this goal, he added.
While Verizon sources in the past had speculated that WDM–PON might provide a pathway toward a reduction in the number of central offices (COs) an FTTH network might require, O'Byrne was cool to the technology. “I believe in order for Verizon to actually deploy it, it would have to resolve some issues,” he said. These would include the potential requirement to provide temperature control for the arrayed waveguide gratings that would provide the WDM function in the field—“Part of our philosophy is to have everything passive in the outside plant,” O'Byrne said—as well as meeting Verizon's requirement for RF overlay support. “We don't have any plans to remove that,” O'Byrne stated, “and that's one of the items which isn't supported by the WDM technologies that we've looked at.”
As an alternative to a pure GPON approach, O'Byrne said that Verizon's GPON vendors, Alcatel–Lucent (www.alcatel–lucent.com) and Motorola (www.motorola.com), have discussed the idea of a hybrid TDM/WDM approach that would “stack” multiple GPONs via WDM. However, about an hour after O'Byrne finished his address, the CTO for the Fixed Access Division at Alcatel–Lucent, Marcus Weldon, suggested that such an approach is unlikely to be forthcoming in the near future. Two primary obstacles stand in the way, he said. First, it would require a new type of splitter that would combine both wavelength and power. Second, the architecture would require an increase in the optical distribution network power budget, which might make direct upgrade of existing GPONs problematic.
“We just haven't yet seen it be ratified by any operator,” Weldon said of interest in the approach. “In fact, the FSAN argued it should be put in as an annex to the current standard in ITU–T, and they would not standardize that approach as an NG PON.”
Similarly, Weldon also didn't see significant work on standardizing a pure WDM–PON approach as part of NGPON2 in the near term. He suggested that, based on current technology, WDM–PON would be outside the realm of economic feasibility for some time to come—perhaps 2x to 3x more expensive in its initial incarnation than current GPON systems. The cost differential derives from several factors, in particular the fact that carriers would need a WDM laser in the CO for every subscriber, which would add both operations expense (in the form of inventory requirements) and capital expense. Weldon noted that WDM lasers would have to support line widths 20x narrower than current PON devices. “That 20–fold reduction in line width has cost, and that cost is on the order of possibly something like $100 per laser,” he said.
Such pricing may not be a significant issue for applications such as wireless backhaul, Weldon suggested. However, “it is a technology that needs fundamental evolution and innovation to get to residential–type pricing,” he concluded.
Weldon had a brighter outlook on other NGPON technologies, however. Weldon expressed confidence that FSAN and the ITU–T would ensure the synergy between 10G GPON and EPON that O'Byrne wants to see. The use of similar IC technology and wavelengths (see the figure) will be significant steps in this direction, which will enable PON lasers to be produced in price–reducing volumes.
Meanwhile, NGPON1 will also see a new specification for more powerful lasers. Dubbed C+, the specification will call for an output power of 32 dB. The extra power could be used to increase reach from the current 20 km to 30 km or double the number of splits supported. While the desire for extra reach initially targeted remote users, Weldon said the current interest in C+ derives from a desire to consolidate COs. The use of C+ optics could enable carriers to reduce the number of COs in a network by 60% to 70%, he suggested.
Making the shunning of WDM–PON nearly unanimous, two of the three participants in an afternoon roundtable on EPON, GPON, and Active Ethernet/point to point agreed that WDM–PON will have trouble establishing itself as a viable option for FTTH in the next few years.
“What it really comes down to is the cost to do that,” offered David Gaetani, director, global customer marketing, access networks solutions, at Motorola. “There may be certain places where it's very applicable. However, I still think it's still quite a few years out; we're talking three to four years out before we see anything that's ready for deployment.”
Shane Eleniak, vice president, marketing and business development, at EPON systems developer Alloptic (www.alloptic.com), had an even dimmer view of WDM–PON. “I'm still looking for a change in technology on how that's going to happen,” he said when asked about making WDM–PON an economically viable option for FTTH. “We looked very closely at that a couple of years ago; we continued to look very closely at it when we tried to decide what our next–generation PON would be. And I think as we look at where can we take it...¦the key driver is going to be what is it going to take to make it cost–effective? And I guess I'm still not seeing it from a technology perspective.”
Yet WDM–PON did have a defender in Wolfgang Fischer, director, business development, Cisco Europe (www.cisco.com). “I think the technology is already reasonably mature,” Fischer said of WDM–PON, which he asserted provides many of the same advantages as the point–to–point architectures Cisco favors for FTTH. He conceded that current WDM–PON technology requires a round or two of cost reduction to bring it more in line with current EPON and GPON costs when measured on a customer–by–customer basis.
“But I would claim that it is a lot more future proof,” he said of WDM–PON. “And when you do the calculation on cost per gigabit per second, I think then already today a WDM–PON solution would certainly be more interesting than a GPON solution. Because when you have 32 [Gigabit Ethernet streams] over a single fiber, that's a lot of bandwidth.”
"Verizon's FTTP Broadband Architecture and Perspective on NG Optical Access,” Vincent O'Byrne, Optical Access ‘08
“Next Generation PON Standards and Technology Update,” Marcus Weldon, Optical Access ‘08
“Roundtable: EPON vs. GPON vs. Active Ethernet: Is the Debate Still Relevant?” Shane Eleniak, Wolfgang Fischer, and David Gaetani, Optical Access ‘08