Active Ethernet FTTH offers PON alternative

by Stephen Hardy

Given all the attention that PON has received, one could be forgiven for the belief that it is the only way to deploy FTTH. Yet hundreds of networks around the world—and particularly in Europe—use Ethernet switches to deliver high-speed voice, data, and video services to single-family homes and apartment complexes. And while incumbent carriers almost unanimously have favored PON, Ethernet proponents say the demands for higher bandwidth plus the opportunities provided by upgrades to fiber to the node and other DSL-based networks means the door to such customers hasn't closed completely.

The use of Ethernet switches in FTTH networks isn't new, even though the approach hasn't garnered as much publicity as PON. One reason for the comparative lack of attention might be the fact that there isn't an agreement among proponents of the technology about what to call it. "Active Ethernet" has gained some favor, as it differentiates the use of powered Ethernet switches in the outside plant from the passive splitters in a PON.

However, not all implementations place the switches in the field; the Ethernet equipment can reside in the central office (CO), with a fiber running directly from the CO to each subscriber. For this reason, Alcatel-Lucent (www.alcatel-lucent.com) favors the use of Active Ethernet to describe an architecture with switches in the field and "point to point" for one with equipment only in the CO. However, PacketFront (www.packetfront.com) doesn't see the need to differentiate and refers to both as Active Ethernet. Cisco (www.cisco.com), the market leader, also doesn't see the need to split hairs; however, it coined its own term, "Ethernet FTTH," to describe what it provides for these applications.

Market research firm Infonetics Research (www.infonetics.com) has adopted Cisco's terminology. Infonetics reports that, whatever you call it, use of such equipment is growing; sales of Ethernet FTTH gear rose 89% in 2007 versus 2006, the researchers estimate. Meanwhile, Dittberner Associates (www.dittberner.com) early this year also forecasted an increase in Ethernet port shipments for FTTH in 2008, from 172,000 ports in 2007 to 575,000 this year. However, that pales in comparison to the dominance of PON; despite the increase in port shipments, Dittberner estimates that Ethernet's market share will plummet from 19% in 2007 to 5% this year as GPON takes off.

But while incumbent carrier purchases will lead to the growth in GPON and other PON sales, Ethernet has plenty of FTTH fans elsewhere. The technology has proven particularly popular among utilities and municipalities—and since these network builders were the FTTH pioneers in Europe, Ethernet currently enjoys the deployment lead there.

Several factors make Ethernet appealing. One is the general familiarity of the protocol. "It feels right to them as a LAN technology. A simple Ethernet technology resonates more strongly with municipalities and utilities than what is perceived as a telco technology, meaning PON," says Marcus Weldon, CTO of the Fixed Access Division at Alcatel-Lucent.

Utilities also have experience with powered elements in the field, making them less wary of putting an Ethernet switch in the outside plant if that should be necessary. "If you look at a lot of PacketFront fiber-to-the-home clients in the EMEA theater, whether they're utilities or alternative operators, they've been building active networks today," according to Tim Scott, PacketFront's vice president of sales. "So the first thing is that there's that confidence in the technology and confidence in the standards and confidence that it works."

Ethernet is also seen as a better method to support an open access services model, which many municipalities and utilities favor. Since a fiber is dedicated to each subscriber from the switch—regardless of whether the switch is in the field or in the CO—moving a subscriber from one provider to another is relatively straightforward. Conversely, it's "pretty tough to do open access on a PON," according to Ian Hood, senior marketing manager at Cisco.

The direct link from the switch to the subscriber offers other advantages as well. First, it enables Ethernet networks to provide the highest bandwidth per subscriber. The amount of bandwidth per subscriber can be adjusted more easily than in PONs, and quality of service and service-level agreement support are more clearly understood and easily implemented when providing business services with the same infrastructure. Encryption isn't necessary either, since the bandwidth and lines aren't shared.

But Ethernet has disadvantages versus PONs, too. The principal shortcoming is cost, whether it comes in the form of having to power an Ethernet switch in the outside plant or, in a point-to-point architecture, running a fiber per subscriber all the way back to the CO.

"Point-to-point, actually, we generally find a harder solution to justify because of all those fibers running back all the way to the central office," says Weldon, whose company, of course, offers PON equipment as well. "So you need large ducts, more rights of way, and you have power and space requirements in the central office that are much more significant because you're bringing back a fiber for every subscriber."

For these reasons, Weldon suggests, point-to-point networks work more economically than PONs only in applications where CO requirements are small and subscriber numbers are low.

Needless to say, other Ethernet vendors disagree. "With every customer that we've worked with—and we've worked with quite a lot of them—they all did a comparison of...how much it would cost us with PON and how much it would cost us with Active Ethernet.' At the end of the day it does depend on the architecture and there were some differences between the customers. But cost-wise, it was very comparable: They talked about the same prices with Active and PON," says Irit Gillath, vice president, product line management at Telco Systems (www.telco.com), another vendor active in FTTH.

Fiber costs are a relatively small percentage of a total deployment, Hood suggests. Also, as bandwidth per subscriber reaches 20 Mbits/sec—
a relatively near-term goal for many carriers—the cost differences shrink as well, Hood says. Bandwidth levels of this magnitude generally require that the number of splits per PON shrink. This translates into more fibers in the field and more ports in the CO to service the same number of subscribers, making PON infrastructures look more like point-to-point when it comes to fibers deployed and CO ports.

"What I see in the marketplace is that providers are putting in PON for I'll call it the low end of the market—best effort, small customers, small businesses, the less than 15 Mbits per customer kind of speeds," Hood says. "And then for their high-value customers and some of their businesses, they're going with an Ethernet solution, be it building-oriented or dedicated from the CO to get beyond the 25-Mbit realm. So you're seeing a hybrid approach basically in a lot of the new large city deployments."

However, Hood concedes that point-to-point Ethernet has limitations. If fiber access is an issue, point-to-point architectures become problematic. Similarly, if bandwidth to each subscriber will be relatively low, than the shared aspects of PON win out economically.

Hood's vision of multiple network technologies is shared by others Lightwave contacted. Scott revealed that PacketFront plans to introduce a GPON offering in the near future. "Active Ethernet has worked for many of our clients today," he says. "But again, some of those clients are now also wanting to complement what they've done with Active Ethernet with GPON—for instance, perhaps in some of the rural areas. Some of the utilities that we have that have rural serving areas, where a GPON solution might make more sense."

On the other hand, Weldon believes Alcatel-Lucent's vision of Active Ethernet will find favor as a fiber-to-the-node extension strategy. "Normally, what we've seen is that FTTN advocates start deploying fiber to the home as an Active Ethernet type solution from a DSLAM, or at least they have a strong interest in it," he explains. "You could argue that in tough economic times, fiber to the node might become more favorable. And fiber to the node leads to some Active Ethernet solutions being deployed perhaps more widely than would otherwise have been the case."

Conversely, the European regulatory insistence on unbundling and/or duct sharing will force incumbents to find economical and efficient ways to support open access initiatives with PONs. Weldon says such steps are already in place, which would make PON potentially more appealing in deployments for the open access business model.

Yet one advantage Ethernet technology currently has should remain: the highest potential bandwidth per subscriber. Even technologies meant to increase PON bandwidth could be bent to Ethernet's uses. Take WDM PON, for example. "If we can get the optics costs down and embed them along with the multiplexer into our Ethernet switches, then you can go beyond 100 megabits to gigabits to whatever you can run on a lambda at that kind of cost for the next evolution of speed," Hood says.