Having embraced fiber-optic technology at an early stage, the MSO has positioned itself well to offer a variety of new services.
We gave you a glimpse last month of what AT&T Broadband (formerly TCI) has planned for the brave new world of broadband convergence. Since Time Warner is the other member of the 10-million-plus cable subscriber club, we thought it useful to examine its network strategy.
Although Time Warner is planning the same types of new services as AT&T, the two companies' cable holdings and their approach to building networks have significant differences. One of the main differences between the two multiple-system operators (MSOs) is the extent to which their systems have been "clustered." This, in turn, has a major impact on how fiber optics and electronics are deployed in their networks.
Clustering is the practice of grouping individual cable networks within logical geographical areas for economies of scale and better market cohesiveness. This practice was first conceived to allow the advertisement insertion business of cable companies to flourish.
With the proliferation of fiber optics and the launch of new digital services in the '90s clustering has become even more important. Among its benefits is the ability to deploy relatively inexpensive fiber-optic ring architectures for transport and shorter distribution fiber routes.
Historically, Time Warner's systems have been much more highly clustered than TCI's, making it better positioned to exploit these benefits. More recently, TCI and now AT&T have been involved in numerous joint ventures, acquisitions, and swaps aimed at consolidating its cable holdings into more tightly concentrated clusters. So, whereas Time Warner has been aggressively deploying fiber rings in its clusters since the first half of the decade, AT&T, overall, is at an earlier stage in this process.
Time Warner has been using fiber-optic ring-ring-star-coaxial-bus architectures since the early nineties. Since the mold was set before the advent of dense wavelength-division multiplexing (DWDM) and the company was committed to build a fiber-optic infrastructure that would not be obsolete anytime soon, the result is rings with relatively large fiber capacity.Primary and secondary fiber rings are built with a minimum of 36 fibers. Typical counts are 48 fibers or even higher when the local operating company has an alternative access business. This amount allows Time Warner to transmit individual services (e.g., video and high-speed data) on separate fibers throughout the ring and to use cost-effective OC-3 or OC-12 circuits with no immediate need for DWDM at least until video-on-demand (VOD) is deployed (see Figure 1). In contrast, AT&T's architecture makes heavy use of DWDM.
A related difference between the two companies' network philosophies is the way distribution hubs are utilized. According to Ron Boyer, a senior corporate staff engineer at Time Warner, the company believes that moving a portion of the server capacity out to the hubs can increase the reliability and flexibility of the network. As much as 80% of movie rentals are for movies in the current top-10 list. Boyer says distributing these 10 top movies to servers in the distribution hubs in advance can significantly lower VOD traffic in the transport network. This technique also sidesteps many of the latency issues inherent in centralized processing.
The technical component of Time Warner's digital-video strategy has been developed in the company's Pegasus project, which is the offspring of Time Warner's Full Service Network (FSN) project in Orlando, FL. Although a success in many respects, the FSN system was too expensive to be widely deployed in its original form. Based on what was learned in Orlando, Pegasus was designed as a trimmed-down approach that makes economic sense while retaining most of the functionality of the FSN.
The Pegasus converter stands as a milestone in the cable industry, since unlike other more proprietary approaches, it was designed based on open technical standards and therefore can be built by any manufacturer. Thus far, Scientific Atlanta and Pioneer have delivered Pegasus set-tops to Time Warner, and the MSO is deploying both vendors' set-tops in some markets. In late summer, Pegasus deployments were underway in 25 Time Warner Divisions with the rest to follow by the end of 2000. Also in the works are VOD trials with Pegasus in the Tampa Bay and Austin Divisions.
Time Warner's VOD plans heavily depend on DWDM. Even with distributed servers, the total bandwidth needed for this service is daunting. For example, if 25% of 300 customers in a 500-homes-passed node service area were watching a VOD movie, you would need approximately 225 Mbits/sec of bandwidth to serve that node. When we consider that Time Warner configures an average of 40 nodes fed from each distribution hub, it adds up to 9 Gbits/sec of VOD data service. Distribution hubs are positioned around a fiber-optic ring from a headend in groups of three to five. Assuming an average of four hubs, 36 Gbits/sec of bandwidth would be needed to support this VOD demand.
Time Warner is planning to package all of the VOD channels initially into groups of eight QAM channels that are channel-mapped between 562 and 600 MHz. Then by using five colors of an eight-channel DWDM system, it will deliver 400 video streams (10 video streams by eight channels by five colors) to each distribution hub. The three unused colors will be reserved for future capacity. At the hub, each color will be brought back to radio frequency, filtered, combined with the broadcast and data signals, and sent to the node using 1310-nm analog transmitters. As the demand for the service grows, Time Warner will move part of the server capacity to take advantage of the fact that demand tends to be highly concentrated among current hit movies.
Time Warner is also making plans to deliver voice-over-Internet protocol (VoIP). The company has had a long-term trial of a circuit-switched telephone service running in Rochester, NY. Although the Rochester trial has been quite successful technically, it has not been expanded due to business-related concerns.
According to Boyer, Time Warner believes VoIP can address many of the business issues surrounding voice service but still faces technical issues such as quality of service that need to be resolved. Another reason Time Warner has held back on a residential voice offering has been its pending deal with AT&T to deliver VoIP. Although the verdict is not officially in on that deal, its closing deadline has come and gone. Once this issue is completely resolved, Time Warner has plans to start one or more VoIP trials-in the near future.
Road Runner, Time Warner's branded high-speed data service, has been widely deployed in the company's cable systems. The successful launch of this service is evidence that hybrid fiber/coax architectures provide a flexible platform to support new applications as they are introduced in the future.Even with highly clustered networks, Time Warner has a small percentage of systems in which its ring-ring-star-bus architecture does not work well. An example would be a small suburb within 20 to 40 mi of a much larger city but isolated by mountains, water, or farmland. Building a ring and distribution hub to deliver new services to these communities would be financially and physically difficult. Boyer suggests that processing these services at the nearest supportable headend or hub and relying on microwave for a redundant path may be the answer. In the reverse direction, a frequency-stacking scheme will be used in most cases.
Mitch Shapiro has been tracking and analyzing the broadband industry for more than 12 years. He is currently a consultant with Pangrac & Associates and author of a series of P&A reports on consolidation, clustering, network upgrades, and new service strategies in the cable industry. He can be reached at [email protected] or via the P&A Website at http://broadbandfuture.com.