Case Study: Fiber-Deep, 5 Years Later

By Jeff Gould, Suddenlink, and Joseph P. Lanza, Aurora Networks

In 2001, the cable system in Malvern, AR, about 50 miles southwest of Little Rock, was in need of a complete network upgrade. The system in place was an early 1980s design: 330 MHz delivering 40 analog TV channels. It was definitely time to prepare this system for the start of the third millennium and the demands of delivering triple-play services.

Objectives

The operations and engineering teams managing the Malvern system wanted to rebuild their cable system cost effectively using a state-of-the-art architecture. Most systems being totally rebuilt during this time used an HFC architecture, designed to 860 MHz with node-service areas of approximately 500 homes, and node plus four to six amplifiers to achieve the necessary reach.

To accomplish their "state of the art" objective, the team in charge of the Malvern rebuild decided to take a hard look at a new HFC architecture called fiber-deep HFC (also referred to as "node plus zero amplifiers"). It was claimed that this fiber-deep HFC architecture would achieve equal or better forward network performance, be able to provide more narrowcasting bandwidth per home passed, utilize a high performance, all-digital return technology for improved return network performance and operating stability, and improve network reliability - at a similar total cost to a traditional HFC network.

Additionally, there would be a saving in operating expenses. Although these expenses are always hard to quantify at the initial stage of a project, the early indicators were good: Active network elements would be reduced by 83 percent (and power supplies by 67 percent), minimizing the need for return system "sweep and balance" and reducing overall maintenance with the fiber-deep deployment. Table 1 provides a comparison of key parameters between an 860 MHz traditional HFC design and a fiber-deep HFC design, along with the actual statistics of the fiber-deep deployment. As Table 1 indicates, Malvern's cable plant passes just under 8,000 homes that are fairly well spread out, an average of only 51 homes passed per mile.

TABLE 1: 860 MHz design summary

During the decision-making process, with the help of Aurora Networks, Malvern reviewed and double-checked detailed comparisons of every aspect of plant construction and related cost assumptions, to make sure nothing had been overlooked. When the process was completed, the engineering and operations teams were satisfied that the numbers were correct; a fiber-deep HFC architecture could be built for the same cost as a traditional HFC network and with long term operating cost savings. When coupled with the increased bandwidth per subscriber, improved system performance, network availability and reliability, Malvern's teams unanimously concluded that fiber-deep was the architecture to deploy.

Design and construction

During the final design and initial construction phases, Malvern personnel encountered a few characteristics of the fiber-deep architecture that required some adaptations to their basic concepts. The most significant of these were:

• The rural nature of Malvern resulted in installation of 41 line extenders to reach homes in the really low density areas of the system. (Malvern did not set a number as a cut-off point. Each area was considered independently. The number of homes passed, the possibility of future growth, and the absence of reasonable alternate routes were used as criteria for the decision.)

• A unique powering scheme of interconnecting feeder lines with power inserters provided power-system continuity to adjacent nodes while blocking the RF passbands.

• The fiber-deep design required field deployment of optical passives. (Malvern had always installed optical passives in the headend.)

• The use of digital return technology enabled concatenation (or daisy-chaining) of the return transmitters from the nodes - typically about six. This is shown in Figure 1.

FIGURE 1: Fiber efficiency using Aurora's digital return technology

The low number of power supplies was a pleasant surprise. The network's bandwidth almost tripled (from 330 MHz to 870 MHz), but the number of power supplies did not increase; it stayed at 20. Malvern did take advantage of the total rebuild to increase the system's operating voltage from 60 to 90 VAC. However, to maintain a true "apples-to-apples" comparison, we calculated that a traditional HFC design would have required 65 power supplies at 90 VAC.

Progressing through the Malvern rebuild, we found that the small-sized node service areas (averaging 50 to 60 homes passed) made it possible to transfer customers from the old system to the fiber-deep system with minimal interruptions to service. This cut-over could have been a customer-relations nightmare. However, with advance customer notification of our plans, pre-certifying the nodes, the speed at which we were able to make the cut-over, and the limited number of customers affected within each node-service area, the transition was very manageable and customer friendly.

Results

As a result of the rebuild, Malvern's 3,000 plus video customers are now being offered 65 analog channels and 154 digital channels (of which 23 are high definition, HD) at a dramatically improved level of picture quality. In addition to the expansion of our video offering, the rebuild of the network enabled us to launch high-speed data and voice services. Our 1,000-plus high-speed data customers are enjoying an average data rate of more than 8 Mbps down and are extremely pleased with our service compared to the alternatives.

The rebuild of Malvern was completed in the summer of 2003. The cost to build the system was almost 10 percent lower on a per-home passed basis than the final fiber-deep cost projections. The single biggest reason for this accomplishment was that the number of homes passed increased from 6,600 to 7,600 (a 15 percent increase) without any measurable increase in plant miles. This result was further confirmed by the fact that the actual cost-per-mile was only 1 percent higher than estimated, although the cost of system passives and taps increased by more than 20 percent.

The rebuild of Malvern went extremely smoothly; we were within budget, on schedule (once we negotiated a few issues with our local power company), had few if any customer complaints during the transition, and our customers were thrilled with the increased service offerings. With training from Aurora, the equipment was easy to install and activate, and it performed as specified. The proofing of the system turned out to be a snap because of its simplicity: very few active devices in the field (1.1 RF actives per mile), minimal system sweeping of the very rural areas (where we used some line extenders), and the digital return was essentially a "set and forget" deal.

Analysis

Our first fiber-deep deployment in Malvern, which turned out to be the first of many, has now been operational for five years. We thought it would be a good time to see if we really did achieve the opex and reliability benefits that we intuitively thought we would.

Once we upgraded Malvern, we put in place a maintenance plan that called for a monthly check of the first and last nodes of each daisy-chain in each of the 17 fiber-deep clusters (each cluster consisted of an average of eight fiber-deep nodes on one return fiber), along with all 20 power supplies. This maintenance procedure takes one individual about six or seven days (four days for the nodes; two to three days for the power supplies) per month. However, system personnel believe the monthly node checks are overkill because, in five years, only one change has ever been made to a node.

In addition to our scheduled maintenance, non-scheduled maintenance has been reduced to almost zero. We now experience very few outages of any kind in Malvern. When we do experience one of these rare outages, the combination of the small-node service areas of the fiber-deep architecture and Aurora's embedded monitoring system makes for quick and accurate troubleshooting - all of which have resulted in a reduction of the time the network is unavailable to our customers (approaching zero), a better result than our most optimistic projections. As further proof of the performance of the Malvern network and its operating team, Malvern received Suddenlink's 2007 Award for the lowest trouble-call percentage per revenue generating unit (RGU), an award we were very proud to receive.

Since we chose to rebuild Malvern with the fiber-deep architecture, it is not possible to directly compare the power savings vs. traditional HFC, but the fact that the traditional HFC design would have required 65 power supplies (90 VAC) or three times the number we actually deployed, we are saving a tremendous amount of money on system powering. Today, each home passed in the Malvern system is costing Suddenlink $1.78 compared to the $4-plus seen in some of the company's other systems that have similar homes-per-mile densities and traditional HFC architecture. The low power consumption of the network is a major operational advantage that will continue to benefit us in the future, especially given recent increases in energy costs.

In our initial reviews of the first architectural design comparisons provided by Aurora, we were a little dubious - more narrowcasting bandwidth, 75 percent fewer actives, improved network reliability and availability, lower power consumption and reduced maintenance, all for about the same capital cost as a traditional HFC plant. After five years of experience in Malvern, we've come to realize that the benefits of fiber-deep architecture are real. We would never want to go back to traditional HFC.

Jeff Gould is vice president of Technical Operations for Suddenlink. Reach him at Jeff.Gould@Suddenlink.com. Joseph P. Lanza is director of Field Services for Aurora Networks. Reach him at JLanza@Aurora.com.