Multiple system operators deploy fiber-ring networks
Mitch Shapiro
Contributing Editor
As the cable-TV industry continues to undergo consolidation in anticipation of head-to-head competition with telephone companies, the industry`s large multiple system operators are beginning to deploy large metropolitan area networks that can support a full range of video, voice and data services. These networks are being built around fiber backbones, which are increasingly taking the form of redundant rings that can provide the high levels of reliability required to become a full-service network provider.
Farther into the distribution network, the trend is toward smaller optical serving areas and fewer radio-frequency active components, a movement also driven largely by reliability concerns.
As they prepare to offer their customers telephone and interactive services, MSOs are beginning to deploy more return-path optical components.
These and related trends are driving continued growth in optical sales to the cable-TV industry. They are also raising new questions about which technologies are best suited to satisfy the industry`s changing needs. Following are snapshots of fiber-deployment plans among the major multiple system operators.
Tele-Communications Inc.
Tele-Communications Inc., the nation`s largest cable operator, has pursued an ambitious construction agenda during the past few years and, according to vice president of engineering, Tony Werner, the company`s pace of construction for 1995 will remain "fairly aggressive."
He says TCI is deploying 750-megahert¥technology in all of its upgrades and rebuilds. He describes the typical architecture as a dual-ring, star bus, noting that it will be scalable in design, with opto-electronics added as demand grows for new services.
At the highest level, TCI networks will consist of optical rings that link hubs serving areas with 60,000 to 100,000 homes. In TCI`s larger markets, says Werner, the rings might link as many as seven or eight of these primary hubs.
These large metropolitan rings will contain "enough fiber to do a variety of things," he says. Depending upon the system`s "immediate needs," synchronous optical network, or a combination of Sonet and a proprietary digital system will be used to transport signals around the primary ring, adds Werner.
Feeding off these rings will be secondary optical rings linking Optical Transition Nodes that each serve as many as 20,000 homes. In markets where TCI or Teleport Communications Group--which it partly owns--is active in the competitive access business, these secondary rings may also be equipped with some Sonet gear. Downstream video signals will be delivered from primary hubs to OTNs in analog mode.
TCI is currently using mostly 1310-nanometer distributed feedback lasers for primary hub-to-OTN links. In addition to "broadcast" channels, which are delivered to all homes, the secondary rings will also use block converters to transmit "narrowcast" digital signals.
At the OTN, 200-MH¥digital narrowcast tiers will be separated and sent to distribution nodes using lower-cost digital-grade analog distributed feedback lasers. An OTN might feed 67 nodes in a star configuration, with each node serving approximately 300 homes via coaxial-cable distribution legs.
Long term, TCI is likely to employ 1310/1550-nm wavelength-division multiplexing, which could cut in half the number of active fibers required in the secondary ring to support narrowcasting. The use of narrow WDM, which would further increase the number of optical transmission windows, would mean that even less active fibers would be needed for narrowcast services.
TCI`s current plan is to install enough fibers--including fibers needed for broadcast and narrowcast video, telephony and spares--to each secondary hub. Opto-electronics for narrowcast signals, says Werner, will be deployed incrementally as they make economic sense.
As with the downstream signals, TCI plans to use block conversion and WDM to get the most bang for every buck it invests in fiber. According to Werner, upstream signals will be carried in the 5-to-40 MH¥band on the coaxial portion of the network as well as on the node-to-OTN fiber links. At the OTN, these upstream signals will be stacked using block converters and WDM.
Initially, says Werner, the 5-to-40 MH¥band on the node-to-OTN optical legs should be able to handle return-path traffic on a single fiber. As the penetration of two-way services grows, additional return-lasers and receivers will be added at the node and OTN, respectively.
Block conversion and signal-stacking will not be employed at this level of the network, according to Werner. One reason, he explains, is to keep "failure groups" sufficiently small.
If all return traffic from a 300-home optical serving area was stacked on to a single laser, he says, a failure at either end of that optical link would knock out service to all 300 homes. If, however, each of three coaxial distribution legs had its own dedicated fiber for return signals, the size of the failure group would be limited to 100 homes.
By combining small failure groups in the distribution plant with ring redundancy at higher levels of the network, TCI intends to provide the enhanced reliability needed to become a full-service network provider.
Time Warner
Time Warner plans to have upgraded 87% of its roughly 125,000 plant miles with hybrid fiber/coaxial-cable architectures by the end of 1998. It is also in the process of acquiring the cable systems owned by two other MSOs--Cablevision Industries and KBLCOM--although it has yet to decide on specific upgrade plans for these new systems.
Upon completion of these acquisitions, Time Warner will have approximately 11 million subscribers, which will make it roughly the same size as the nation`s largest MSO, Tele-Communications Inc.
Most of Time Warner`s planned upgrades will provide 750 MH¥of bandwidth and extend fiber nodes to areas of roughly 500 homes.
Once the test is over, says Don Gall, senior project engineer, the multiple system operator may begin using the concept in other markets. Although the most likely prospects are markets with higher densities, he says the architecture "might be cost-effective down to densities of 90 homes per mile." Companywide, Time Warner`s average density is roughly 94 homes per mile.
Gall says the fiber-deep architecture is likely to deploy six fibers to each 500-home area, where an optical splitter will be used to feed two to four optical receiver nodes. Each node, he says, would be followed by no more than three line extenders in the coaxial portion of the network.
Like other top MSOs, Time Warner has aggressive plans to deliver telephone and data services and narrowcast and interactive video services.
Although it is currently not planning to employ redundancy in the residential distribution plant, Time Warner is planning to do so at higher levels of the network.
This will typically include dual-rotating 36-fiber rings that link 3 to 5 distribution hubs, each serving an average of 20,000 homes. Each hub would, in turn, be optically linked to approximately 40 fiber nodes, which each feed coaxial-cable distribution legs serving an average of 500 homes per node.
In the MSO`s larger markets--in Orlando, FL, for example, Time Warner serves roughly 1 million subscribers--these rings may be interconnected by a 48-fiber metro-area transport ring. While none of these larger rings have been deployed yet, some are in the active planning stage.
The current plan is for analog broadcast channels to be carried on the rings via 1550-nm high powered lasers, perhaps amplified by erbium-doped fiber amplifiers. Digital narrowcast services, carried in the 550-to-750 MH¥spectrum, would be transported on other fibers in the ring via Sonet.
At the hubs, the 200 MH¥of digital narrowcast signals would be dropped off the Sonet ring and modulated onto 6-MH¥carriers using 45-megabit-per-second QAM modulators. Although Time Warner`s Orlando full-service network trial is using 23 QAM channels per node, Gall suggests that commercial deployments in other markets may start with only a few modulators per node and add more as demand warrants.
With each hub feeding approximately 40 nodes, a fully equipped hub might eventually house 900 QAM modulators. To plan for this, Time Warner, while today building hub sites containing 100 square feet of space, is planning for expansion to 600 to 1,200 square feet, says Gall.
The company is also exploring an even more aggressive "fiber-deep" architecture, which would split optical signals and extend fiber deeper into the plant so that nodes would serve only approximately 125 homes. To test the fiber-deep concept, Time Warner has undertaken a 200-mile test of the new architecture in a Pittsburgh suburb.
Cox Cable Communications Inc.
Cox serves roughly 1.8 million subscribers and is adding another 1.2 million via purchase of Times Mirror`s cable systems. It currently operates 51,000 miles of coaxial plant and 4,000 route miles of fiber, with an average of 35 fibers per route mile.
By the end of 1998, Cox plans to serve 95% of its customers with a hybrid fiber/coaxial-cable network that provides at least 550 MH¥of bandwidth. More than half of its customers at that point will be served by systems equipped for 750-MH¥operation.
The company has plans to deliver telephone service in most of its major markets and has allied itself with TCI, Comcast and Sprint in the provision of nationwide wired and wireless telephony. Thanks to a Federal Communications Commission-granted Pioneers Preference, Cox has also been granted a license to provide wireless personal communication service in a large area that includes southern California and Las Vegas.
To prepare itself as a full-service network provider, Cox recently began deploying a new hybrid fiber/coaxial-cable architecture to provide improved reliability, flexibility and capabilities. This new approach, referred to as "ring-in-ring," links nodes serving 1,000 or fewer homes via optical rings.
According to Alex Best, senior vice president of engineering, Cox plans to deploy the architecture in all its existing systems--many of which are in fairly large metropolitan markets. This marks the first time a cable MSO has committed to bring ring-redundancy that deep into its distribution network. The extent to which it will be deployed in Times Mirror systems--some of which are fairly small--is under consideration.
The ring-in-ring architecture is designed to deliver three general classes of service--"lifeline," which requires adherence to the Bell Communications Research standard of less than 53 minutes of downtime per year; "broadcast," which includes all services that originate from the headend and are distributed uniformly among all customers; and "targeted" services, which are intended for specific customers or groups of customers served by particular optical nodes.
The new architecture`s name reflects its design. It comprises two rings--a dedicated ring in which fibers originate at the headend and run to each node over diverse routes; and a second continuous loop-through ring that passes through each node. The dedicated ring delivers broadcast and targeted services from the headend to each node. It may also be used to transport upstream signals from the node back to the headend.
The dedicated ring will consist of four fibers, two each for downstream and upstream traffic. If, in the future, additional nodes or services are added to the network, two of these fibers could be reassigned to handle this expansion and wavelength-division multiplexing or bidirectional couplers would be employed so both upstream and downstream traffic could be carried on a single fiber.
The loop-through fiber would also include four fibers, two for telephony plus two spares. The loop-through fiber can also be used as a cost-effective backup for the broadcast video signals in case there is a failure in the dedicated loop.
In the future, if traffic on the dedicated ring becomes congested, broadcast video signals could be transferred to one of the loop-through fibers, notes Best. These signals could be sent over the ring and dropped off at each node, using a high-powered transmitter and, if more power is needed, an erbium-doped fiber amplifier.
Best says the cost of a ring-in-ring architecture--not including the extra transmitters required for route diversity--is roughly equivalent to that of a star-based network that delivers eight fibers to nodes that serve 1,000 homes. Whereas the ring approach involves more route miles, he explains, a star configuration requires more fiber miles.
Diverse-route transmitters will be added when Cox begins offering telephone service, explains Best. They may also be fired up for video services if the competition requires it.
Thus far, according to Best, Cox has activated the return optical path in very few systems, typically for the low-data rate links required for impulse pay-per-view. This will change, he says, as telephony, high-speed data services and interactive video are introduced.
Although 1310-nm distributed-feedback lasers account for the majority of transmitters deployed by Cox, Best says the company has used some high-power YAGs to replace micro wave links.
Even though Cox has yet to deploy any 1550-nm optical technology, Best says this is being considered, probably in conjunction with EDFAs, for headend-to-headend interconnects. He sees increased need for such links because Times Mirror systems are integrated into Cox`s existing system clusters.
Others looking at ring-in-ring
Although Cox appears, for now, to be the only MSO embracing the idea of rings in the distribution portion of the network, other large MSOs have begun to evaluate the ring-in-ring approach.
Continental`s senior vice president of engineering and technology, David Fellows, began his analysis of the ring-in-ring approach with serious questions about whether a ring-based network was the most cost-effective way to achieve high reliability in the distribution plant. His inclination had been to focus instead on deployment of highly reliable equipment and on the ability to quickly diagnose and correct network problems.
Fellows also foresees other hurdles a ring-based distribution plant must overcome. In some markets, he suggests, operators will find it difficult to find redundant routes between nodes.
The architecture Adelphia Communications is deploying in most of its major markets resembles Time Warner`s fiber-deep approach. Deployed for the first time in Syracuse, NY, about two years ago, Adelphia`s near passive architecture typically limits fiber serving areas to a few hundred homes and roughly two miles of coaxial plant. The result is much smaller failure groups.
In Alexandria, VA, Jones Intercable is also deploying a near-passive hybrid fiber/coaxial-cable network. This project is not typical, however, says Ken Wright, a recent member of Jones` corporate engineering group until he accepted a position with Intermedia Partners.
The Alexandria system upgrade, which will limit optical serving areas to no more than 200 homes, was justified by a number of factors unique to Alexandria, says Wright. These include upscale demographics, high PC penetration and household densities, and the impending launch of competition, in the form of Bell Atlantic`s video-on-demand service.
Wright says Jones Intercable, which plans to double in size with the help of a large equity investment from Bell Canada, is taking a market-driven approach to network upgrades. For each market, it will deploy the architecture that makes the most economic sense in light of that city`s density, demographics, PC penetration, overall market demand and level of competition.
While top MSOs serving larger markets are installing 750-MH¥plant almost exclusively, MSOs such as Cablevision Industries and Post-Newsweek, which tend to serve smaller markets, are deploying mostly 550-MH¥plant. (In the case of Cablevision, this may change because the company is being acquired by Time Warner).
The number of homes served by a fiber node also tends to be larger in systems serving smaller markets. MSOs such as Falcon and Century, for example, are involved in upgrades that leave as many as 10 radio-frequency amplifiers in cascade and that serve as many as 3,000 to 5,000 homes per node.
These upgrades, which can be completed quickly and at relatively low cost, tend to resemble the upgrades undertaken in larger markets several years ago. And like these, they leave the door open to more aggressive fiber deployments when this is justified by emerging business opportunities.
Mitch Shapiro is a freelance writer based in Encinitas, CA.