Are PONs too small for the big fish?
The prospects for fiber in the U.S. local loop brightened recently when the Federal Communications Commission handed the incumbent local-exchange carriers (ILECs) a virtual monopoly on broadband services, provided that those services reach customers over new fiber-optic infrastructure. As ILECs provided much of the input to the first International Telecommunication Union (ITU) passive-optical-network (PON) standards (Recommendations G.983.1 and G.983.2) as part of the Full Service Access Network (FSAN) initiative, it seems reasonable to conclude that (assuming the ILECs take the FCC's hint) the promise of PONs in the United States may finally come to fruition in the relative near-term.
This belief gains further credence in light of new standards that increase the capacity of the average PON. However, a growing number of observers, including some carriers ready to put fiber into the ground, question whether active technology might not provide a better way to carry broadband services over optical cable.
PONs derive their name from the fact that the optical infrastructure between the optical line terminal (OLT) in the central office (CO) and optical-network terminal (ONT) at the user's premises (or wherever the endpoint of the optical line may be) does not contain active components. Instead, a passive splitter divides the downstream bandwidth —622 Mbits/sec in the original FSAN standards, which supports 155 Mbits/sec per ONT in the opposite direction—among 16 users or more, depending on what variant of PON is deployed.
Because the focus of the original PON standards was to ensure carrier class voice quality while also supporting data and video services, ATM served as the base protocol. "APON [ATM-based PON] takes advantage of the built-in capability of ATM to differentiate between different types of traffic and to apply different types of delay and QoS [quality of service] parameters to them," says Susan Heilman, product manager at Optical Solutions (Minneapolis). Optical Solutions, arguably the most successful vendor of PON equipment for FTTH applications in the United States, bases its "broadband PON" architecture on ATM. "ATM automatically gives POTS or T1 traffic guaranteed throughput and delay that is required to maintain call quality. ATM also recognizes best effort data as a type of traffic that is willing to tolerate some level of delay and queues it and sends it only when bandwidth is available."
But as equipment vendors began to target business applications, the general datacom trend toward Ethernet, coupled with the relatively limited bandwidth afforded by FSAN-based PONs, led to the development of Ethernet PONs (EPONs). Now in the IEEE standards-development process as IEEE 802.3ah, "Ethernet in the First Mile" extends the IEEE 802.3 media access control (MAC) and MAC control sublayers with additional physical layers. The concept also includes a point-to-multipoint architecture similar to FSAN but will provide greater bandwidth: 1.25 Gbits/sec in each direction.
Ethernet lacks a carrier class QoS mechanism, which means that an additional protocol, such as MPLS, may have to be used to satisfy demanding customers. Despite this shortcoming and the fact that the standard won't be ratified until next year, several Ethernet-based systems have already reached the market. Meanwhile, several vendors supporting APONs have developed equipment that increases the supported bit rate or the number of users (sometimes through the introduction of WDM technology) to address areas where the market may feel the FSAN standards fell short.
However, the FSAN/ITU community has realized that user requirements threaten to exceed their original vision. This past January, the ITU ratified a pair of new standards, Recommendations G.984.1 and G.984.2, that create a new class of passive architecture, the Gigabit PON (GPON). The new specifications cover systems with nominal line rates of 1.25 and 2.5 Gbits/sec downstream and 155 Mbits/sec, 622 Mbits/sec, 1.5 Gbits/sec, and 2.5 Gbits/sec from the customer back to the CO. The new specifications implement the generic framing procedure (GFP) to support data (including Ethernet), video, and voice.
The fact the new standards support twice the bit rate of the proposed EPON specifications tells only half the story, according to Gary Lee, chairman of the PON Forum (www.ponforum.org) and chief executive of FlexLight Networks (Kennesaw, GA), a developer of GPON class equipment. "The real issue here is how this bandwidth is actually used for network traffic. Because GPON utilizes the very efficient GFP protocols, a small amount of the total bandwidth has to be utilized for signaling and other overhead. The GPON protocols operate at 93% efficiency and above," he told Lightwave in an e-mailed response to our questions.
The standards also support the services carriers most often provide in a manner to which they are accustomed. "GPON fully supports all Ethernet services that the service providers wish to provision—[virtual] LAN, QoS, CoS [class of service], etc.)—while also supporting native voice services," Lee writes. "DS-1, E1, OC-3, etc., are mapped onto the PON using GFP in their native formats. There is no requirement to encapsulate the traffic into either Ethernet frames or IP (voice over IP), and full clock recovery, jitter, wander, etc., are guaranteed just like in today's SONET networks."
Despite the advancement of PON standards, the technology has had a checkered success rate globally. In Europe and the United States, most PON users include alternative carriers and real estate developers, with incumbent carriers limiting their applications mainly to trials. "The Asian market, in sharp contrast, is actively testing and deploying PON today," asserts Lee. "China, Taiwan, Korea, Singapore, and Japan all currently have PON in either testing or deployment phases."
Furthermore, the FCC's ruling during the recent Triennial Review may kick-start PON deployment by the incumbents in the United States, Optical Solutions' Heilman believes. "The regulatory hurdle had been key, but it has been crossed," she wrote in another e-mail. "Given the clear paths to success that early implementers of PON are on, continuing market and operational success seems imminent."
But not everyone shares Heilman's confidence. "Slow acceptance of PON will continue in the U.S.," reports David Gross, research director, optical networking, at Communications Industry Researchers (Charlottesville, VA). "In the residential market, the problem is not just PON, it's the high cost of building fiber laterals. For business customers, excluding small businesses that hook into residential PONs, PONs' point-to-multipoint topology is a poor fit for their transparent LANs and internal MANs. Carriers selling to these businesses have to provide point-to-point networks, even if routed through a CO."
The new standards, whether GPON or EPON, won't help, Gross believes. "G.983 has been around a few years now, and its ratification did nothing to stimulate demand. The standards body and the FSAN Consortium that worked with it were filled with PTTs and large vendors focused very heavily on engineering and not on business requirements. Reducing transceiver costs through a passive architecture does little good when labor costs to construct fiber laterals are rising," he said via e-mail, adding, "The IEEE 802.3ah Task Force is not going to help much by standardizing EPON, because no one is rejecting PON because of the Layer 2 framing protocol."
The bottom line, according to Gross, is that "FSAN and the ITU thought they could sell to every market sector, and that clearly is not going to happen."
By way of example, LBDC FTTH Holdings (New York City) recently opened a tender for an FTTH and Wi-Fi network based on an active topology. According to Neal S. Lachman, president/CEO at LBDC and executive director of the new 21st Century Infrastructure Consortium (see "FTTH industry group founded for the little guy," page 30), LBDC favors using an active network. "There is a maximum of 32 connections per PON infrastructure, which is one of the main reasons why I find this technology a very expensive solution," he wrote in an article offered to Lightwave for use in this report. "PON infrastructures are also not futureproof in the sense that PONs are affected by technological limitations today, thus requiring companies to invest sooner rather than later in order to keep abreast of the technological advancements of other entrants in the FTTx market."
LBDC has come up with an active architecture it calls FiberBroadband, which the company believes will enable it to provide lower connection costs than standard PONs while maintaining the ability to boost data rates and features as customer requirements change. The architecture, which is also point-to-multipoint, features a switch in place of the splitter. The customer interface unit is a set-top box rather than a larger, more expensive ONT.
Wave7 Optics (Alpharetta, GA) offers equipment based on a similar approach. According to the company's president/CEO, Tom Tighe, Wave7 targeted the cable TV and multiple-service-operator (MSO) market with an architecture similar to the hybrid fiber/coax topologies those companies had already installed. The company offers systems that create an IP-based optical Ethernet network that will support voice, video, and data. The equipment includes a "Last Mile Core" device that can be located up to 44 mi from a network ring or headend; a "Last Mile Tap" that can be located up to 6 mi from the core device; and a "Last Mile Gateway" that serves as the customer premises device. One core device can support up to 96 gateways and provide up to 500 Mbits/sec in burst capacity per subscriber.
Tighe thinks it makes more sense to put a powered device at the neighborhood distribution point—where other powered telecommunications devices are likely to be located—and reduce the power requirements at the customer's location, rather than follow a PON approach. The vendor has convinced 16 companies in the United States and two internationally of the wisdom of this approach.
Regardless of the technology chosen, optical access infrastructures won't become ubiquitous until the incumbent carriers in the United States and Europe follow the lead of their Asian counterparts. Needless to say, vendors and analysts active in this space differ in their timetables for the advent of this development.
"PON deployments by ILECs and municipalities are ramping up significantly," writes Optical Solutions' Heilman. "RBOCs were hesitant to deploy PON, because it was not clear whether or not they would have to provide their competitors with access to their optical networks. The FCC has now determined that RBOCs will not have to open these networks to competitors, so that roadblock at the highest level has been removed. RBOCs can now freely include PON in their growth planning strategies."
Adds PON Forum's Lee, "We do see activity beginning in the RBOCs, with BellSouth beginning their FTTB trials this quarter, and Verizon due to release an FTTH/FTTB RFx that will be based on PON. We believe PON will be a major technology as the carriers begin spending their capex [capital-expenditure] budgets again."
Yet, Gross doesn't foresee a spike in incumbent PON activity in the near future, at least not in the United States. "Even with the recent removal of unbundling requirements on fiber links, the RBOCs are not going to do much with it beyond a few showcase projects like SBC Mission Bay," he writes. "There is a reason that it is called the 'last mile problem.' There is no magic solution that will solve it."
That said, Gross foresees increased support from PTTs, which should bode well for PONs worldwide. And independent carriers, utilities, and municipalities—the technology's current primary market—appear sold on the approach. The slow creep of fiber closer to end users worldwide has indeed begun.
Claiming that the FTTH Council "actively lobbies for the monopolist incumbent local-exchange carriers" to the detriment of "smaller-sized FTTH companies," a fiber-to-the-home (FTTH) startup has formed a new promotion and trade organization. The new group, 21st Century Infrastructure Consortium (21st CiC) will "create a platform for FTTH builders, operators, and service providers; promote, stimulate, and assist them; form partnerships with suppliers, vendors, consultants, engineers, and technology providers; and introduce industry players with solution providers, content owners, and advanced service providers," according to organizers.
21st CiC hopes to become for the global FTTH industry what the National Telecommunications Cooperative Association (NTCA) is for cable TV operators, in the words of a recent press release. "Being in the FTTH business for almost three years, with no progress of any kind in this brand-new industry, we found it necessary to stir up the FTTH world. With all due respect to all members of the FTTH Council, we need action, not just empty words and promises. We needed an organization that creates opportunities, not only hopes for it," says Neal S. Lachman, executive director of the new consortium and president/CEO of LBDC FTTH Holdings (New York City).
Given the opportunity to respond to Lachman's charges, Donna Keegan, executive director of the FTTH Council, wrote in an e-mail, "The FTTH Council actively supports the deployment of FTTH by all providers and believes any entity should have the right to deploy FTTH. This is reflected in our membership and in our policy positions. The FTTH Council welcomes other groups and individuals to assist in the education of the benefits of FTTH."
The self-proclaimed "world's most ambitious FTTH startup," LBDC says it was the driving force behind the creation of 21st CIC. The consortium will target seven "groups of interest": FTTH/FTTD (fiber-to-the-desk) system operators, non-cable TV system operators, competitive local-exchange carriers (CLECs), TV networks, associates (manufacturers and/or sales firms with hardware or software other than programming for use in FTTH/FTTD Internet service providers [ISPs] or TV distribution systems), affiliates (companies that provide services to FTTH/FTTD ISPs or TV distribution systems), and promotional organizations. But in keeping with its affinity with small companies, it will not allow ILECs, CLECs in which ILECs have a 25% or greater stake, or cable TV operators with more than 100,000 subscriber connections. Cable TV operators controlled by a larger operator or multiple service operator will be similarly barred from joining.
Membership is free to individuals. Information on various levels of corporate membership is available on the organization's Website, www.21stCenturyInfra.com.