Archives

February 2001 Issue
Feature: Deep Fiber Networks
A Review of Ready-to-Deploy Architectures
By and

Pushing fiber deeper may provide a solution for increasing capacity and reusing bandwidth.

In hybrid fiber/coax (HFC) networks, all services are ostensibly "broadcast" with the prime difference being the footprint over which these services are broadcast.

Channel lineups for broadcast video services typically cover the largest area. Advertising zones usually are second on the order of a common 20,000-home hub. For initial penetrations of high-speed data services, such as cable modems, a typical hub site divides into several sectors using a single 6 MHz channel.

Telephony services are broadcast over the smallest area—typically a 6 MHz channel for each node. As penetration of these services increases, the broadcast area for each also will decrease.

Video: Multiple customer types

As operators deploy interactive digital set-top boxes and digital-ready TV sets become available, the types of customers served by multiple system operators (MSOs) will increase. Today’s MSO typically only has to concern itself with two types of subscribers—those with set-top boxes and those without. But as digital service rollouts expand, the types of customers will increase. Some subscribers will have analog TV sets and some will have digital. Some will have analog set-tops, others will have digital and still others will have no set-top at all. And because cable TV is a fairly high-penetration service, MSOs still will need to offer video services to all types of customers.

This point impacts the engineering community in several ways.

First, even after the rollout of digital services, large numbers of purely analog subscribers will remain, accustomed to 80 to 96 channels of analog programming. It would prove difficult to remove those channels for use as digital programming.

Second, pending "must-carry" rules for new digital TV (DTV) and high definition TV (HDTV), programming also could require MSOs to provide these channels.

Finally, over-the-air digital programming will exist in a multilevel vestigial sideband (VSB) format while HFC channels from the headend are modulated via quadrature amplitude modulation (QAM) formats. Operators may need to simulcast some digital broadcast channels to satisfy the needs of these two types of subscribers.

The net of these changes is that broadcast services will eat a large amount of spectral channel allocation, leaving less room for advanced interactive services.

Data and voice: Two-way bandwidth

As operators deploy data and voice services over the next few years and their penetration grows, the need to allocate capacity and hence bandwidth to these services also will increase.

For network simplicity reasons, most MSOs allocate one (or at most two) 6 MHz channel each for cable modems and telephony. The area over which each of these services broadcasts will have to decrease as capacity needs increase.

The developing use and maturity of the Internet further exacerbate additional capacity needs for cable modems. Current cable modem deployments allow for a minimum available data rate of about 128 kilobits per second (kbps) per user. While this rate is a large improvement over current telephony-based modems, the availability of downloadable video clips will pressure MSOs to dramatically increase this minimum available data rate.

For cable telephony, several providers are discussing plans to offer as many as four telephony lines per subscriber, with at least one of them as an "always-on" data channel. For systems allocating one telephony channel per typical 500 home node, the logical node size (the area over which the service will be broadcast) will have to be reduced even further.

Figure 1 illustrates these constraints. The inclusion of large numbers of analog channels for non-set-top subscribers, HDTV channels in possible must-carry scenarios, advanced services such as near-video-on-demand (NVOD) and video-on-demand (VOD), data, voice and other services may push the required bandwidth in excess of 1 GHz. Because most systems are upgrading to 750 MHz and possibly 870 MHz, something will have to occur to accommodate these demands.

One solution involves allocating less bandwidth to the interactive channels and reusing this spectrum over and over again by reducing the logical node size for the interactive channels. From the standpoint of the reverse channels, similar problems exist, resulting in the need for smaller logical node sizes. In addition, for ingress and channel redundancy issues, not all of the 5 MHz to 40 MHz bandwidth is available. Figure 2 shows the available reusable bandwidth as the node size decreases from 500 homes to 50 homes passed in a typical 80 to 100 home-per-mile plant. Experts envision that nearly 2 megabits per second (Mbps) per home passed within 10 years. If we consider optimistic projections for service rollouts and penetrations, this number could expand greatly.

Advanced services and reliability

Increased network reliability proves a primary consideration. Two ways to increase the reliability of a network are to reduce the amount of coaxial cable in the plant along with the number of active devices in the field. Figure 3 illustrates the number of active devices such as RF amplifiers as the physical node size decreases. A minimum in the number of actives, and therefore the maximizes plant reliability, is achieved at the passive cable point (the point at which no additional actives are needed). As the node size is reduced further, additional fiber-optic receivers are needed and the number of actives increases. Also note in Figure 3, that the amount of power needed per mile of plant is shown.

Deep fiber architectures

A deep fiber architecture is similar to a standard HFC network, but contains important differences.

First, typical HFC networks are based on logical node sizes in the 1,000- to 2,000-home range. As logical node sizes approach the passive HFC point, the amount of independent (nonbroadcast) information delivered per mile of plant increases dramatically. For this to occur, the processing equipment at the hubs, or the dense wavelength division multiplexing (DWDM) technology required to remote the hub equipment to the headend, must become more extensive.

Second, the increase in the number of fibers employed in these deep fiber architectures must be handled by in-field splitting or by creating smaller hub sizes.

Finally, driving fiber deeper in the network reduces the numbers of RF amplifiers in cascade so the performance required by each of the elements in the network may be re-optimized to this new configuration in both the forward and reverse.

Deep fiber technology

Because deep fiber architectures employ more optoelectronic components than standard HFC architectures, the steep improvements in both the cost and performance of optoelectronic devices have had a major impact on pushing fiber deeper. Optical transmitters and receivers, as well as fiber optic passives such as splitters and multiplexers, have seen major improvements in terms of performance, reliability and cost.

DWDM technology and high-power optical amplifiers

The introduction of 1,550 nanometers (nm) transmission technology with high-power optical amplifiers, DWDM transmission and WDM overlay technology also largely impacts the deployment of deep fiber systems.

Figure 4 shows the progression of the cost of transmitting light at 1,310 nm and at 1,550 nm. As with all laser-based systems, the costs of generating light per milliwat (mW), year after year, are the result of technology learning curve improvements. Not only does 1,550 nm transmission technology have a much steeper learning curve, but it also moves to higher powers and hence lowers price-per-milliwatt transmission costs.

Because in nearly all HFC networks the majority of the traffic remains broadcast, designers may create systems that take advantage of the low costs afforded by 1,550 nm high-power optical technology. For narrowcast traffic, DWDM transmission and overlay technology allow the narrowcast portion of the spectrum to be transmitted separately, allowing each portion of the transmission to be independently optimized.

At the headend, the quadrature amplitude modulated interactive narrowcast channels are mapped to an optical wavelength that is transmitted to the hub via DWDM at the headend. At the hub, the narrowcast DWDM channels are routed optically to their respective node or group of nodes, where the narrowcast and broadcast portions of the spectrum are combined optically on a single photodetector. The different RF spectrums of the broadcast and narrowcast traffic allows this accomplishment without interference.

Performance improvements

In deep fiber HFC networks, up to 10 times as many nodes exist as in a traditional 500-home node network.

At first glance, one would assume that the optical transmission system would have to supply 10 times the optical power for the broadcast traffic to maintain adequate carrier-to-noise (C/N) ratios.

Figure 5 shows this to be otherwise. As fiber goes deeper into the network, the RF amplifier cascade is eliminated. The specifications usually reserved for the fiber-optic node become the same as the end-of-line specifications.

In this configuration, the optical modulation index of the 1,550 nm transmitter may increase so that the fiber optic link performance matches the end-of-line requirement.

With this, the optical power required at the node to maintain adequate C/N ratios may fall substantially. In fact, for an order of magnitude increase in the number of nodes in a deep fiber network, only 3 dB extra fiber-optic broadcast optical power is required.

Deep fiber economics

Despite the advantages gained in terms of network capacity, performance and reliability when fiber is pushed closer to the home, the high cost of such systems has proved the major stumbling block.

When most operators began their plant upgrade plans a couple of years ago, 500-home nodes were the limit for how deep fiber could go in an HFC network. Improvements in optoelectronic components, new transmission techniques and reduced cost of optical fiber have caused this fiber limit to push closer to the home.

Figure 5 illustrates these improvements. The premium for driving fiber to the 125-home near-passive HFC architecture is less than 10 percent of the cost of rebuilding a 500-home HFC network.

For passive networks, the premium is about 40 percent. Looking at the graph, it appears that about three quarters of the cost is in fiber and construction, while the contribution from electronics cost grows as fiber goes below the near passive point. As the optics and electronics costs continue to fall, experts see the cost of passive HFC networks dropping to levels close to current HFC plant costs.

Donald Sipes is vice president of advanced technology and networks, optoelectronics division, at Scientific-Atlanta. You may reach him at . Bob Loveless is S-A’s director of strategic planning and technology, transmission network systems. You may reach him at .

 Back to February 2001 Issue

Access Intelligence's CABLE GROUP
Communications Technology | CableFAX Daily | CableFAX's CableWORLD | CT's Pipeline
CableFAX Magazine | CableFAX databriefs | Broadband Leaders Retreat | CableFAX Leaders Retreat