PON architectures open the door to business over fiber

June 1, 2000

Jim Holley
NEC Eluminant Technologies Inc.

Exciting new ways of bringing fiber closer to the small and medium-sized business are developing rapidly, which is a necessary follow-on to large corporations increasing the use of fiber optics throughout their networks.

While today's telecommunications network is clearly a mish-mash of copper, fiber, coaxial cable, wireless, and microwave, there is a growing realization that fiber can deliver higher bandwidth, while other media eventually will fall short. And if a carrier can't provide the fiber-based access network that these businesses require, another carrier will.

Carriers must consider a variety of criteria when building-out networks. Higher bandwidth capacity and the ability to deliver voice, video, and data over the same medium are at the top of most lists. Increasingly, carriers recognize that fiber optics represents the most definitive answer to solving these two key challenges.

Today, network build-outs happen in two separate directions. While national backbone construction has brought prodigious amounts of fiber to city borders, fiber is also coming in to the distribution portion of the network. At the other end, large buildings and corporate campuses are now routinely wired with fiber. What's missing is a bridging technology that can eliminate the bottleneck between the end user and central office. In this type of application, a passive-optical-network (PON) architecture makes sense.

PON architectures have been around for about a decade. The PON model has had strong support from leading carriers such as Deutsche Telekom, France Telecom, and NTT for a number of years. Deutsche Telekom, through its Optical Access Lines (OPAL) project, wired some 1.2 million homes with fiber in the former East Germany. The total cost to Deutsche Telekom was $26 billion.

This, and other efforts, led to a standards initiative-the Full-Services Access Network, or FSAN-created in 1995 by 20 major carriers, including Deutsche Telekom, France Telecom, NTT, BellSouth, SBC, British Telecom, and a host of others. FSAN validated the PON architecture and gave it definition (see Table).

A PON is a single-fiber, singlemode system. Its main component is an optical splitter device that splits incoming light and distributes it to up to 64 locations. Architectures were designed for various fiber-to-the-x (FTTX) office, home, cabinet, and curb applications.

Why has PON received such support? For one thing, it has been a sound point-to-multipoint architecture that can segment the advantages of an optical fiber as an almost limitless source of bandwidth by bringing numerous signals on and off a fiber.

A PON can also keep costs down by eliminating outside-plant components, relying instead only on the system's endpoints for active electronics. The endpoints are the optical line terminal (OLT) on one end and optical network termination (ONT) at the other.

The FSAN model uses a double-star architecture with the first star at the OLT, where the wide-area-network interface to services is split and switched to the PON interface. The second star occurs at the splitter, where information is passively split and delivered to each ONT.

Typically located in the carrier's central office, the OLT sends signals to the edge of the network and serves as an interface point between the access system and the service points in the carrier's network. The OLT behaves like an Asynchronous Transfer Mode (ATM) edge switch with ATM-PON interfaces on the subscriber side and ATM-Synchronous Optical Network (SONET) interfaces on the network side.

The integration of a PON with an existing telecommunications architecture is relatively straightforward. As broadband services begin to show up more in the network, that is especially true. The infrastructure for fiber with PON is remarkably similar to copper (asymmetrical digital subscriber line or ADSL) and wireless (local multipoint distribution system). All are using ATM for service delivery, meaning that a service management infrastructure installed for ADSL will work just as well for fiber and wireless.

Large companies can often gain immediate access to fiber because of greater capacity demands such as OC-12 (622-Mbit/sec) or OC-48 (2.5-Gbit/sec) lines. Smaller companies, particularly those slowly converting to broadband services, may not require more than T-1 (1.544-Mbit/sec) lines.

PON systems allow smaller businesses to link to fiber, displacing copper-based Integrated Services Digital Network (ISDN, 64- and 128-kbit/sec) or T-1 lines. It boils down to economics. There is ample evidence to suggest that the price points for fiber are getting close to-if not equal to-those of copper. The question really is how long it will take fiber to cost out, not whether it will do so.

Even though full-fiber throughputs are not always immediately economically feasible, PON systems work well with other media for the last several hundred feet. (The FSAN standard specifies a variety of FTTX architectures.)

PON products can replace the older, copper-oriented equipment between the central office and end user-equipment that is outliving its usefulness. The solution is not digital-loop carrier (DLC) cabinets stuffed full of electronics at or near capacity. Electronics require maintenance, use batteries that need changing, and are subjected to extremes of hot and cold. In the PON system, DLC cabinets are replaced by a generally maintenance-free 6x1-inch PON.

Bandwidth can also be incrementally provisioned over time as needed. For example, if a small business needs 1-Mbit/sec capacity at first but then requires 2 Mbits/sec in 12 months, the carrier only has to provision higher ATM PVC (permanent virtual-circuit) rates, rather than ordering more T-1 lines.

Such systems allow small businesses to deal with the bursts of bandwidth that new services such as video might generate. Requirements can be easily met by remotely provisioning the capacity without truck rolls. The result is a system the company can grow with as part of a "broadband-by-the-drink" approach. There is also peace of mind knowing that the carrier has futureproofed the network for its customers.

Incumbent local-exchange carriers (ILECs) face competition as never before from cable-TV companies, competitive local-exchange carriers (CLECs), and long-distance companies entering the local exchange.

Fiber-rich cable-TV companies generally are seen as leading the broadband race with 1.7 million subscribers, dwarfing the few hundred thousand consumers that telecommunications carriers have been able to attract, according to Forrester Research. Thus, ILECs may want to consider deployment of fiber-based access networks now rather than later.

In addition to analysts' recommendations, there are other signs of progress. Last year, BellSouth replaced copper with fiber-to-the-curb (FTTC) to service about 500,000 homes. SBC has embarked on a $6-billion build-out that is expected to bring far more fiber into its network. Bell Atlantic has reached an agreement with Metromedia Fiber Network that will bring fiber to many businesses and closer to more residences.

BellSouth, which is working cooperatively with NTT, has also announced plans to install a fiber-to-the-home (FTTH) system in Atlanta using an FSAN-compliant PON technology, the first such trial in North America. SBC is expected to consider implementing PON systems.

The question is whether the response will be strong enough, soon enough. Businesses of all sizes will continue to clamor for bandwidth. A PON gives ILECs, and other carriers, the opportunity to complete the network in a way that is advantageous to customers. While the cost of fiber continues to be a concern, the advantages of installation now outweigh the temporary economics of waiting, particularly if fiber will eventually be required anyway.

NTT has installed PONs primarily to outfit large corporate customers in cities like Tokyo and Osaka for several years now. The Japan-based carrier has a qualification process in place for FSAN-compliance and several vendors are now believed to be certified. While the plans of NTT and the Japanese government to build a nationwide FTTX system have slowed somewhat, the business portion of that build-out has begun in earnest, where it economically makes sense to do so.

British Telecom deploys what it calls its "narrowband predecessor" of PON, known as TPON (technology over passive optical network) in some portions of its network. The carrier is considering the use of ATM-PON as a broadband successor as well as alternative solutions based on existing metallic lines and DSL technologies.

In the United States, the ability of a PON system to interface with existing protocols is a major plus. Ethernet and T-1 lines represent examples of what can be transported over an ATM-PON. All legacy services and future services can be readily transported.

That does not mean the FSAN PON model is not being Americanized to fit this unique market space. It is. NEC has put a lot of effort into trying to build a PON system for the American market, taking what it has learned in Japan, but also recognizing that a variety of changes need to be implemented.

The widespread use of multimode fiber-primarily for in-building applications-in the United States is one clear area of difference. This approach is at odds with the singlemode fiber used in the FSAN model generally deployed in Japan. The FSAN PON model also uses a single fiber for both the upstream and downstream paths. Two wavelengths are used-1,550 nm for downstream and 1,310 nm for the upstream-similar to what is used in hybrid fiber/coaxial cable-TV analog networks. In the United States, a converter is placed between the multimode fiber-or some other medium-and the singlemode-fiber network outside the building.

In the United States, an intense struggle between copper and any kind of fiber for close-in applications continues to focus on the economics. That is expected to change as bandwidth requirements become more acute and PON as well as other close-in fiber architectures reach wider acceptance. Even the most diehard multimode-fiber proponent would acknowledge that singlemode fiber is ultimately a better medium, all other factors being equal.

Whatever the final details look like, PON appears destined to play a significant role as optical transport finds its way to or near the business and residence. That is especially true as PON products reach mass production levels.

Jim Holley is senior vice president of marketing at NEC Eluminant Technologies Inc. (Herndon, VA).

  • 1982 Invention of passive optical networks (PONs) at British Telecom labs.
  • 1987 Early PON trials at Bishop's Stortford, UK.
  • 1993 Beginning of massive Deutsche Telekom installation of PON architecture in eastern Germany.
  • 1995 Formation of the full-services access-network group, known as FSAN.
  • 1998 NTT begins installation of FSAN in Japan; the network includes FSAN-compliant products from vendors participating in the FSAN initiative.
  • 1999 BellSouth completes beta testing of PON architecture to 400 homes in the Atlanta area.
  • 1999 International Telecommunciation Union (ITU) adopts FSAN specifications as recommendation G983.1.