Comparing Integrated Broadband Architectures

Comparing Integrated Broadband Architectures from an Economic and Public Policy Perspective

Nosa Omoigui*

Marvin Sirbu**

Charles Eldering***

Nageen Himayat****

published in Telecommunications and Internet Policy, Brock, G., ed. (Lawrence Erlbaum: Washington, DC, 1996).

Abstract

The potential demand for new telecommunications services and the desire by telephone and cable companies to provide them invariably leads to the question of which architecture is best suited to provide these services. Much prior research deals with the engineering and economic merits of different architectures--particularly Hybrid Fiber-Coax and Fiber-in-the-Loop approaches. In this article, we also treat the economics of the different architectures, but, in addition, we place emphasis on the public policy issues that arise when one considers which architecture to deploy. In particular, we compare HFC and FITL architectures from an economic and policy standpoint.

Introduction

Current residential telecommunications services are limited to telephony in the form of Plain Old Telephony Service (POTS), narrowband ISDN and broadcast analog video services. In the near term, video conferencing will create a demand for telecommunications services at rates of n x 64 kb/s or using new virtual circuit technologies such as ATM. Future video services will include broadcast digital video and video-dialtone services.

Of all the various architectures that have been considered for provision of such services, Hybrid Fiber-Coax (HFC) and Fiber-to-the-Curb (FTTC) are the main contenders. HFC is very popular in the cable industry and is currently used to provide analog broadcast services. It will also be used to provide digital broadcast and interactive services in the future. The FTTC system is an embellishment of the Digital Loop Carrier (DLC) architecture which traditionally was largely used by RBOCs to provide telephony services. In the FTTC system, an Optical Network Unit (ONU) is used to serve a small number of subscribers (typically between 8 and 32) using a fiber connection from the Central Office. FTTC is being seriously considered among the telephone companies.

Description of the Basic Architectures

In this section, we describe the different architectures being considered for the provision of broadband interactive services. These architectures can be further categorized into two levels of sophistication: the plain vanilla flavors that provide only telephony or analog video services, and the upgrades to these same architectures designed to provide interactive video and other advanced services.

Architectures for Narrowband or Broadcast Services

This section focuses on three baseline architectures designed for telephony or broadcast video service provision: Hybrid Fiber-Coax (HFC), Digital Loop Carrier (DLC), and Fiber-to-the-Curb (FTTC).

HFC Architecture

The HFC architecture, as the name implies, employs a combination of both fiber and coaxial cable. HFC employs a fiber backbone which corresponds to the fiber feeder in the DLC and the FTTC networks. HFC can be used to deliver both distributed video and telephone services, and can be made to carry up to 500 channels (for a 500 home node, this implies one channel per home). In addition, HFC systems can simultaneously support Video-on-Demand (VOD), and telephony. For broadcast transmission, HFC provides each subscriber with the same group of video channels-- therefore, there is high sharing of resources. The HFC headend receives signals from local studios, over-the-air broadcasts, or microwave and satellite sources and then combines and re-transmits these signals over the trunk (or feeder) cable of the network. Portions of the signal are then split to feeder cables and then in turn to drop cables to serve the household. The HFC distribution plant consists mainly of coaxial cable. Because of the large attenuation of signals over coaxial cable, feeder amplifiers (mini-bridgers and line extenders) are necessary to amplify the signal for both forward and return paths. Passives include taps, connectors, 3-way splitters, power inserters, and directional couplers. Sixty volt power supplies supply power needed to drive the actives in the distribution network. The drop loop consists of coaxial drop cable. Lastly, an addressable converter is needed on the subscriber premises to deliver distributed and switched video services.

The HFC system is a shared transport medium; signals for all the subscribers fed from a given node are available at every household served by the node. This requires encryption to prevent unwarranted reception by some subscribers. In the return path, signals from all households are summed at the node and transmitted to the headend. The fact that all signals from the subscriber locations are received at the headend results in the phenomena called "noise funneling" in which unwanted signals (noise and ingress) from all the subscriber locations appear at the headend. This requires the use of a multiple access technique such as TDMA, FDMA or CDMA to allow the various subscribers to transmit to the headend. The use of fiber-optic transmission systems from the headend to the node has resulted in a decrease in node size, with the typical node size today being in the range of 500-2000 homes passed. At node sizes of 500, the ingress problems are greatly reduced, and the amount of spectrum available in the downstream and upstream would support narrowband services at full penetration rates as well as interactive video services.

Figure 1 shows the main components of an HFC architecture.

Figure 1: Diagram Showing Main Components of the HFC Architecture

DLC Architecture

Digital Loop Carrier (DLC) is a fiber-based architecture which is typically used by telephone companies to provide narrowband telephony service (POTS) [1]. The DLC system analyzed for this paper has terminals both in the CO and in the RT. The system employs the SONET signal format as the transmission medium between terminals. Each terminal can support two independent directions of SONET transmission, and consists of three major subsystems: a common control assembly (CCA) --optical receive and transmit, modulation and demodulation of SONET carrier, and SONET formatting. Each common control assembly can support up to nine channel bank assemblies at the OC-3 rate, each of which supports up to 56 channel cards. Each channel card can support up to 4 lines; therefore each terminal can support up to 2016 lines. The channel bank assembly houses the various channel units and channel bank common equipment units. The DLC distribution plant is composed of twisted pair copper cable of different sizes, with appropriate connectors, splices, and pedestals. The drop loop includes pedestals at the curb, and copper drop cable. Subscriber Premises Equipment includes a protective block for surge protection. Figure 2 below shows the main components of a DLC architecture.

Figure 2: Diagram Showing Main Components of the DLC Architecture

FTTC Architecture

The FTTC architecture is very similar to a DLC system. In an FTTC system, fiber is extended past the node, and to the pedestal--which is at the curb, near the home [1]. This has several advantages. First, it exploits the high capacity of fiber by sharing the transmission facilities over more subscribers in the distribution network Second, by deploying more fiber, it lowers the incremental investment needed for broadband services. The drop from the curb is assumed to be twisted copper pair, although if FTTC is to be used to provide video, a dual drop cable containing both telephony twisted pair and coax may be cost effective in comparison with the cost of the electronics needed to multiplex and demultiplex telephony and video bandwidth signals over a twisted copper pair drop alone. In an FTTC system, remote electronics--referred to as a Host Digital Terminal--takes a high bit rate signal from the Central Office Terminal and demultiplexes it via a fiber bank assembly onto lower bit rate distribution fibers. The HDT may also contain a Network Power Assembly (NPA) to provide power to the Optical Network Units (ONU) located at the curb. When power comes from an NPA, the distribution fiber cable must include a copper power feed.. Fiber bank assemblies provide the optical transmit, receive, and multiplexing functions for the optical link to the ONUs. The ONU contains the channel units or line cards which interface a customer's copper wire based equipment to the terminal. The FTTC distribution plant is composed of fiber cable of different sizes again including splicing and connector costs. The drop loop and subscriber premises equipment are essentially the same as in DLC.

This narrowband architecture is well suited for reliable delivery of POTS, ISDN, and special services and is well-positioned to meet the higher bit rate demands for remote LAN access and video-telephony applications. A key difference between the HFC and the FTTC architectures is the means of transporting digital signals from the Remote Terminal to the curb FTTC uses baseband transport of digital signals over low noise fiber; HFC requires the modulation of an RF carrier with the digital signal for carriage over coax from the Remote Node. Both systems use copper for the drop. In an FTTC system, the choice of copper twisted pair for the drop creates more noise exposure than the use of coax drops which are the norm in an HFC system.

The main components of an FTTC architecture are shown below in Figure 3.

Figure 3: Diagram Showing Main Components of the FTTC Architecture

Architectures for Broadband and Interactive Services

Modifications to the baseline infrastructures described above are required if these architectures are to support the full range of services encompassing analog broadcast, interactive video and telecommunications services. These extended architectures are now described.

Advanced HFC

One approach to these advanced networks is to upgrade the basic HFC infrastructure. An Advanced HFC (AHFC) architecture is shown in Figure 4. Additional equipment is now required at the headend to support interactive video and telecommunication services. The HDT must be capable of interfacing to an ATM network as well as a Public Switched Telephone Network (PSTN) to accept video and telecommunications traffic. The digital data streams are then combined with broadcast signals and switched to individual subscribers requiring service. Subscribers communicate with the headend via settop boxes for low bit rate interactivity suitable for video dial tone applications and return telecommunications traffic is coupled to the return channel through a premises interface device termed the coaxial termination unit, which is placed on the side of the residence.

From a technical perspective, the challenges to this approach include managing signal security via access control and encryption as well as overcoming impairments on the return channel for providing highly reliable service required by telecommunications applications. However, there is no fundamental reason why the HFC network cannot be upgraded to provide the full range of telecommunications services. As will be shown, the HFC network is an extremely cost-effective transport medium.

Figure 4: Diagram Showing Main Components of the "Advanced HFC" Architecture

FTTC with HFC overlay

Assuming a telephone company plans on installing an FTTC network to support telephony and narrowband switched services, there are then two alternative routes for providing broadcast services: using an overlay HFC network, or by upgrading the FTTC network to

carry video. Typically, an HFC network will carry 30-100 channels simultaneously, in analog form. This allows the recipient to use today's Cable Ready TV sets to receive the signal. If the HFC system is installed at the same time as the FTTC system, than the net additional costs of the overlay HFC are quite small, since the two systems can be pulled simultaneously, and a siamese drop cable--with both coax for HFC and twisted pair for FTTC can be pulled for the cost of pulling either one singly. By contrast, providing broadcast services over FTTC is more costly.

Indeed, it is not economical to provide video services in a broadcast mode over FTTC. Wideband lasers capable of carrying broadband signals are very expensive; the alternative of providing digital transport and digital to analog conversion at the ONU for up to 100 broadcast channels is also prohibitively expensive. Instead, the FTTC network must be upgraded with digital ATM switches to provide a Video Dial Tone capability which allows the customer to dial up either broadcast services or VOD. In addition, because FTTC systems carry video in digital form, not analog, the consumer needs a high cost (~$350) settop box to decompress the video signal and convert it to a form suitable for use with analog TV sets. Once these upgrades have been made, however, the FTTC system can provide interactive video services (IVS), or switched video services (SVS)--such as Video on Demand--for no additional cost. By contrast, if the analog HFC overlay route was chosen, additional investment would be required to upgrade either the HFC or the FTTC to carry IVS.

We will consider three options starting with FTTC for telephony:

1. FTTC for telephony and HFC for both broadcast and IVS services (FTTC(T)-HFC)

2. FTTC for telephony, HFC for broadcast and FTTC for IVS services(FTTC-HFC(B))

3. FTTC for telephony, broadcast and IVS services. (All-FTTC--broadcast provided identically to IVS).

The FTTC/HFC hybrid architecture is shown in Figure 5. The HFC-FTTC convergence is accomplished close to the living unit in a combining tap that serves the premises with the necessary physical media.

Figure 5: Diagram Showing Main Components of the "FTTC with HFC Overlay" Architecture

All FTTC

As noted above, an FTTC network could provide switched digital access to today's broadcast and premium analog cable channels as well as to Video On Demand. Switched digital access means that the consumer would be obliged to have a set top box which ordered the appropriate channel to be delivered down the line, and converted the signal from digital to analog form. However, the Cable Act of 1992 directed the FCC to issue rules designed to oblige cable service providers to support the provision of multiplexed analog signals in a form compatible with today's cable ready TV sets.[1] The required interposition of a set top box in an FTTC-only system may require a waiver from these rules if the FTTC system is construed as providing "cable service." Telephone companies desirous of providing "cable service," but, wary of seeking a waiver under the current rules, may believe themselves obligated under the rules to construct an HFC overlay on top of an FTTC system, rather than service all video needs through a single FTTC network as switched digital video, regardless of which method is less expensive. Because of these rules, we are analyzing the all-FTTC option so as to shed light on the actual cost to the telephone companies and to consumers should the FCC refuse to grant the necessary waiver. In the event waivers from the Cable Act mandate are available, than the all-FTTC option must be considered on its economic merits.

DLC with HFC overlay

Like FTTC, DLC can also be upgraded to support interactive services by overlaying it with an HFC network. In this scenario, the HFC network carries broadcast services and interactive video services, while the DLC network carries only telecommunications services. The architecture is similar to that of the FTTC-HFC overlay architecture where HFC is used for both analog and IVS services. Unlike FTTC, it is not possible to carry video directly on the DLC system at acceptable bit rates.[2]Telephony is provided over the DLC portion of the network--which uses copper in the distribution plant and eliminates the ONUs of FTTC.

Economic Models

In this section, we present an economic comparison of HFC and FTTC-based architectures when they are deployed to carry a full range of services. In particular, cost[3] analysis of the HFC architecture supporting analog broadcast, interactive video, as well as telecommunications services is presented.

Scenarios Considered

The economic models developed in this study considered the following scenarios for the deployment of telecommunications and interactive video services:

1. HFC network carries broadcast services and interactive video services (IVS) and DLC network is used for telecommunications services.

2. HFC network carries broadcast and interactive video services and the FTTC network is used for telecommunications services.

3. HFC network carries broadcast services only and interactive video services are placed on the FTTC network in addition to the telecommunications services.[4]

4. All services are carried over the HFC network--in the "Advanced HFC" configuration.

5. All services are carried over the FTTC network.

Cost comparison of these architectures is based on developing accurate cost models for laying out the HFC and the FTTC infrastructures to enable them in their traditional roles of supporting interactive video services over the HFC and the FTTC architectures are determined separately, along with the incremental cost of carrying telecommunications services over the HFC architecture.

The incremental costs are then combined with the infrastructure costs to obtain the overall results.

Assumptions

The set of assumptions governing the costs models are outlined below. These assumptions are representative of current visions of network deployment.

1. The node size is assumed to be 480 homes, along with a central office size of 30,000 homes and a network headend size of 180,000 homes.

2. The plant is assumed to be 70% aerial and 30% underground with a 60% take rate for analog services. For modeling, the cost of the penetration rate for interactive services is set to be a variable assumed to be less than the take rate for analog services.

3. A settop box estimated to cost $125 is needed in conjunction with an HFC network carrying broadcast analog services. A digital settop box, which is needed for all IVS services and for enabling the all-FTTC network to carry broadcast services is estimated to cost $350.

4. The peak coincident "busy hour" usage of these services is considered to be fixed at 25% of the interactive service penetration rate. In addition, it is assumed that no further cabling infrastructure resources are required to support IVS.

5. IVS offerings are considered to be 70% requiring 1.5 MB/s and 30% requiring 6Mb/s, with a settop to user ratio of 1.6. This corresponds to an average bandwidth of 2.85 Mb/s per IVS channel. Note that this imposes a relatively low demand on bandwidth per channel. Demands on system bandwidth may be higher, in general, and would certainly be higher if HDTV services are to be introduced.

For modeling the cost of SDV services over the FTTC architecture, the system proposed by Broadband Technologies (BBT) is used as a reference. Interactive video services over the FTTC architecture supports very high bandwidth in the forward direction (approximately 2.4 Gb/s or an equivalent of 400 channels at 6Mb/s is provided per node). Channels are switched to individual subscribers via ATM switches at the headend. Each subscriber is provided the capability of receiving six video channels. High-bandwidth return capability is provided and the telecommunications traffic is transported in combination with video information.

The FTTC architecture used in this analysis represents a custom development carried out by Broadband Technologies that attempts to optimize ATM switching technology to account for asymmetric bandwidth requirements of IVS in the forward and return directions. Additional optimization is also carried out by combining with video, telecommunications data streams already present on the FTTC networks. From the viewpoint of cost modeling, it becomes difficult to assign costs for a custom developed architecture unless detailed system designs are available. Hence our approach has been to estimate the cost of a system providing SDV functionality equivalent to that of the telecommunications services. We treat telecommunications and video services independently but this assumption is not allowed to affect the cost estimate significantly. This approach is useful in providing a rough estimate of the incremental cost of SDV services over FTTC and as is shown later, is extremely useful in comparing the general trends in cost with respect to both architectures.

The following assumptions are used in the incremental cost analysis of FTTC:

1. A node size of 1000 is chosen; the central office and headend size are assumed the same as for HFC.

2. A 100% penetration rate is assumed for telecommunications services, therefore additional infrastructure equipment is not required for deploying video services.

3. One optical network unit (ONU) is deployed for every 16 subscribers and it is assumed that the IVS subscribers are uniformly distributed amongst all ONU service areas.[5]

In both the HFC and FTTC instances, cost comparison is carried out on a "cost-per-home-passed" basis. A 40% markup is also applied to the incremental costs for interactive video services to account for installation costs. Additionally, for comparing the cost of IVS services over the HFC or FTTC network, a suburban plant (comprising 100 homes per mile passed) is assumed with node sizes being equivalent to those considered for the IVS cost modeling.

Summary of Component Costs and Economic Assumptions

The tables below provide more detail on the underlying component costs assumed in the economic analyses. Economic models were developed using the Analytica engine from Lumina Decision Systems. For certain variables, such as trenching costs, which vary with terrain, we assumed a probabilistic distribution of cost rather than a fixed estimate. The final cost numbers reported in section 4 should thus be understood as the mean value of the resulting probabilistic distribution of total costs Installation costs are based on a sample of contemporary projects. Component costs are based on several manufacturer's list prices, adjusted for estimated quantity discounts.

Cost of analog settop $125
Cost of digital settop $350
Penetration analog video 60%
Number of settops per subscriber 1.6

Table 1: Settop costs and penetration assumptions

Component

Cost Units
Single-mode Fiber 0.2 the number of fibers $/meter
Feeder Fiber Sheath 0.6$/meter
Feeder Inner Duct 1.15 $/meter
Fiber Splice normal(34.4, 11.5) $/splice
Fiber Pigtail 250$/unit
Power Per Watt (WOW) 10 in C.O., 15 in R.T. $/Watt
Fiber Strand Installation cost normal(1108.8, 153.94) $/mile
Fiber Lash Installation cost normal(1963.8, 353.7) $/mile
Fiber Duct Installation cost normal(2227.8, 673.01) $/mile
Trenching Cost normal(14870, 5833.9) $/mile

Table 2: Estimated Costs of Common Network Components

As noted in section 2.1.2, in the DLC architecture, a Remote Terminal consists of a cabinet, channel bank assemblies, common control and power equipment. The total equipment cost of a fully equipped suburban node serving 1,000 homes is $219,000. In the table below we note some of the key component costs that went into calculating the above figure, as well as other key cost assumptions for the DLC models.

Location Component Equipment Cost ($) Installation Cost ($)
Central Office (COT), Remote Node (RT) Optical Transmitter Unit (OTU) 1170 40% markup
Central Office (COT), Remote Node (RT) Optical Receiver Unit (ORU) 1070 40% markup
Central Office (COT), Remote Node (RT) SONET Formatter Unit (SFU) 2520 40% markup
Remote Node (RT) Channel Bank Assemblies $90,000/1,000 homes 40% markup
Remote Node (RT) Line card (included in Channel Bank Assemblies) 118/4 lines 40% markup
Distribution Copper Splice 1.66 n/a
Distribution Copper Connector 20n/a
Distribution, Drop Copper Inner Duct 0.75 ($/meter) n/a
Distribution Copper Cable (Twisted Pairs) (varies by cable size) 1.65 - 10.8 ($/meter) aerial - normal(3.75, 1.2); underground - normal (4.75, 1.5)

Table 3: Table of Estimated Costs of Main DLC Specific Network Components

In the FTTC architecture, fiber-based channel bank assemblies at the Host Digital Terminal - HDT carry signals to an Optical Network Unit (ONU) from which copper drops extend to each household. The HDT may be located either in the field--similar to a DLC RT--or inside the central office. A fully equipped and installed HDT supporting 1000 homes costs $283,000.

Location Component Equipment Cost ($) Installation Cost ($)
Remote Node (HDT) Fiber Bank Assembly 11,550 40% markup
Optical Network Unit (ONU) 24 line ONU with line cards 1,990 40% markup
Remote Node (HDT) Network Power Supply (NPS) 47340% markup
Remote Node (HDT) Network Power Controller (NPC) 17340% markup

Table 4: Table of Estimated Prices of Main FTTC Specific Network Components

Location Component Units Equipment Cost Installation Cost
Drop Loop RJ-6 Drop Cable (Aerial) $/ft0.095 for rural and suburban plants; 0.112 for urban see below
Drop Loop RJ-6 Drop Cable (Underground) $/ft0.084 for rural and suburban plants; 0.117 for urban see below
Distribution Directional Coupler $19.75 25
Distribution Power Inserter $19 25
Distribution 3-Way Splitter $25 25
Distribution Mini-bridger $769 45
Distribution Line Extender $189 35
Distribution 2-Port Tap $7.4 15
Distribution 4-Port Tap $8.53 15
Distribution 8-Port Tap $13.28 15
Distribution 60V Power Supply $1710.28 25
Headend Forward Laser and cabinet $5758 137
Headend Return Path Receiver & Chassis $795 137
Headend Commander Modulator $1300 275
Headend Commander Demodulator $3000 275
Remote Node Forward Costs $2191 375
Remote Node Return Costs $765
Subscriber Premises Settop $125 50
Distribution 750' Coaxial Cable $/ft0.91 for underground, 0.752 for aerial see below
Distribution 500' Coaxial Cable $/ft0.45 for underground, 0.359 for aerial see below
Distribution Aerial Coax Strand Installation $/mile normal(1050.7, 194.38)
Distribution Aerial Coax Standard Install Cable $/mile normal(1322, 9.08)
Distribution Coax Trench Installation $/mile normal(12950, 4191.1)


Table 5: Table of Estimated Costs of Main HFC Specific Network Components

In order to deliver digital video over an HFC system, it must be enhanced with additional equipment for carrying wideband digital signals. Typically a basic carrier rate of 29 Mbps per 6 Mhz video channel is used for carrying video traffic. The number of channels that can be carried on a single 29 Mbps carrier depends upon the compression rate--live action typically requires more bps than a movie which can be pre-processed. HDTV signals consume 4-6 times the bit rate of NTSC signals. A digital carrier totaling 29 Mbps representing 4-10 channels will be transmitted by satellite to the cable headend where an Integrated Transencryption Multiplexor (ITEM) decrypts the signal and then re-encrypts it for transmission over the cable. This digital signal is modulated for transmission over the fiber backbone using a 64 QAM modulator. Return signals from the settop boxes are modulated for transmission upstream from the remote node using a QPSK modulator. A Network Control Unit processes the upstream signals from the settops.

ItemCost
Integrated TransEncryption Multiplexor (ITEM) $50,000
64-QAM Modulator3,500
QPSK Modulator2,500
QPSK Receiver10,000
Network Control unit 30,000

Table 6: Prices for Advanced HFC Digital Video components

For ATM Costs, prices have been falling at 30% per year. These prices are as of early 1995. Also, we assume that the cost of an OC-12 port would be 4 times that of an OC-3 port.

Item Cost
Cost of Fore Systems ASX200 ATM Switch (2.4 Gbps Switch) 21,950
Cost of ATM Output Port 2,500
Cost of ATM Input Port 10,000


Table 7: Estimated Costs of ATM Components ($)

Location \ Architecture DLCFTTC
Feeder Electronics 164.29146.46
Feeder Electronics Labor 65.7258.59
Feeder Cable20.10 20.09
Feeder Cable Labor 31.2431.24
Distribution Electronics 309.10304.15
Distribution Electronics Labor 123.70121.70
Distribution Cable 196.80151.27
Distribution Cable Labor 107.8390.31
Drop Electronics75.00 305.11
Drop Electronics Labor 91.20183.20
Drop Cable60.28 54.09
Drop Cable Labor117.59 224.06
Total Cost1362.85 1690.27


Table 8: Cost Breakdown of Total Cost Per Home Passed (DLC and FTTC) -- 1000 Home Node, Suburban Topology

Results and Economic Comparisons

The results for the analysis can be best understood by dividing the architectures into two categories: baseline architectures that only provide analog broadcast and telecommunication services, and advanced architectures that also provide interactive broadband services.

Baseline Architectures

Figure 6 shows the baseline cost for HFC, FTTC and DLC, respectively. These are not directly comparable of course, since the baseline HFC system provides only broadcast video, while the other two provide only telecommunications. Still, we can note the higher cost for FTTC over DLC to provide basic telephony; this suggests that FTTC can be cost justified only in terms of future services, such as video.

Figure 6. Price per Home Passed for Baseline HFC, DLC, and FTTC networks

Advanced Architectures

In order to provide telephony services, an HFC network must be augmented with additional electronics and premises equipment. These costs have been estimated at $400 per household, with the assumption that 100% of are telephone subscribers. Similarly, in order to provide broadcast video services, an FTTC or DLC network must be augmented by an overlay HFC network. Further investment is needed to bring any of the three architectures to the point where they can handle Switched Digital Video (SVS) services as well.

Figure 7 shows the combined result for the five scenarios outlined in the previous section. It is seen that the most cost-effective solution is to use the HFC architecture to carry all services. This primarily results from the fact that the initial infrastructure costs for HFC are significantly lower when compared with the FTTC/DLC infrastructure costs [1].

Figure 7: Cost Comparison of all Services over Hybrid/Single Architectures

KEY:

From Figure 7, notice that the Advanced HFC architecture is the least costly at all IVS penetration rates between five and fifty percent. This is followed by the HFC-DLC hybrid architecture, where HFC carries interactive video services and DLC carries telecommunication services. Within this same range of penetration rates, the remaining three scenarios have comparable costs. However, we notice that as the take-rate increases, the cost for the HFC-FTTC hybrid architecture (where FTTC carries telecommunication services) crosses that of the all-FTTC and the FTTC-HFC (where HFC carries only analog broadcast services). Indeed, the three lines seem to cross at a penetration rate of 25%. At higher IVS penetration rates, notice that the all-FTTC option is preferred.

We also note that FTTC costs are essentially flat with the penetration of interactive video services (IVS), but HFC costs increase steadily with penetration. This is illustrated in Figure 8 which shows the incremental cost of providing Switched Digital Video (SVS) services over HFC and FTTC architectures. This also implies that at IVS penetration rates higher than 50%, the all-FTTC option would become significantly cheaper than all the hybrid architectures and would eventually become as cheap as the all-HFC architecture.

Figure 8: Comparison of the Incremental Cost of Switched Digital Video (SDV) Services over HFC and FTTC

It is seen that the IVS cost over HFC increases with increasing penetration rate while IVS cost over FTTC is relatively insensitive to the penetration rate. This can be explained by the fact that the FTTC structure provides for a large bandwidth that is fixed regardless of the penetration rate. This implies that fixed costs are incurred for the ATM switch and the ATM switch interfaces to the ATM network. Additionally, the position of the ONUs are fixed by the demand for telecommunications services. Therefore, even if a few subscribers are present in an ONU service area, high-bandwidth output ports in the ATM switch, and the optical units within the ONU are still needed to support them. The only component cost that changes with the penetration rate is the cost of subscribers' video channel cards. Hence, for low penetration rates, IVS over FTTC is more costly than that over HFC but the cost does not changes significantly with increasing penetration rate.

In contrast, the cost of IVS over the HFC network grows as the demand for bandwidth through the system grows. The increasing costs of IVS service over HFC as a function of system bandwidth is shown in Figure 9. Note that these figures do not include settop costs and are shown here primarily to illustrate how incremental costs would rise with increasing bandwidth. The bandwidth of 273 Mbps corresponds to the bandwidth used for the HFC system at 50% penetration rate in Figure 7. The bandwidth of 1.24 Gbps corresponds to the bandwidth that is provided by a SDV system over FTTC for a 500 home node.

Figure 9: Incremental Cost of SDV Services over HFC as a Function of Bandwidth (273.6 Mbps Corresponds to the Bandwidth Requirement at 50% IVS Penetration Rate, 25% Coincidental Usage (96 Channels) and an Average Bandwidth of 2.85 Mbps per Channel)

Policy Implications

The main question that policy-makers and industry analysts have been asking is whether cable companies have a chance to compete with the RBOCs during and after the convergence of their respective media. The results of this analysis (see Figure 7) indicate that cable companies are very well positioned to compete with the telephone companies. Indeed, Figure 7 shows that in cases where there is small to moderate penetration for interactive services (which would likely be the case in the near term), the costs to the cable companies compare extremely favorably to what the RBOCs would have to spend. In order to provide a full range of interactive video services, cable companies need only upgrade their existing HFC broadcast network at modest cost (Figure 8), whereas a LEC must either install an FTTC system or an HFC system from scratch.

The FCC's Video Dial Tone rules--which may be altered should proposed legislation pass--were intended to oblige telephone companies to build Video-Dialtone (VDT) systems as opposed to being allowed to build vanilla cable systems. This policy has been undermined, however, by Appellate Court decisions which have struck down the prohibitions on telephone companies providing "cable service." Thus, telephone companies appear to have a choice between offering "cable service" under Title VI of the Communications Act or VDT service under Title II.

One might want to ask what the impact of the two regimes is on the telephone companies, and in the long-term, on the consumer. The incremental cost of providing VDT service versus plain cable service follows directly from this analysis: it is the difference in cost of an HFC or FTTC system providing switched video services and that of a baseline HFC network providing only analog broadcast video. For an HFC system, adding VDT capabilities can more than double the cost. The FCC justifies the requirement by pointing out were the telephone companies to build systems capable only of carrying analog broadcast channels, they would not be able to meet the common carrier obligation of serving all-comers. However, this requirement means that the RBOCs are being forced to spend a minimum of an additional 100-300 dollars per home passed to build their networks. In addition, the rule is premised on the assumption that consumers would not mind paying more for the added value. In the long run, however, if consumers cannot rationalize paying more for these advanced services, the telephone companies would be at a distinct disadvantage because the cable companies would be able to offer cheaper rates for the plain vanilla services. On the other hand, if the FCC mandate leads telephone companies to install all-FTTC systems, they will be better positioned in the event IVS is highly popular than if they had built a standard HFC system to compete with the cable companies.,

If the recent appellate court decisions are not overturned by the Supreme Court, telephone companies may be permitted to offer "cable service" under Title VI. Should they do so, however, they then become subject to the FCC rules which require that basic tier cable service be receivable on cable-ready TV sets. This latter requirement effectively forces the telephone companies to build HFC overlay systems, if they want to be regulated as cable service providers, even if they build FTTC systems for later provision of Switched Digital Video.[7] The results of our models indicate that the cost of this requirement is significant, and amounts to as much as $300 per home passed.[8] Without this requirement, the case for integrated HFC versus an integrated FTTC system would be much less clear. While an all HFC system still looks most attractive, the price difference is small enough, particularly at high IVS penetrations, that modest improvements in the underlying economics of FTTC, or other considerations, such as long term maintenance costs, could swing an RBOC to choose FTTC over HFC as the architecture of choice.

The analysis also makes clear that FTTC systems cannot be cost-justified to handle voice only. If the RBOCs build voice-only FTTC systems, and the regulators allow all of the costs to be put in their telephone service rate-base, then consumers will be paying a significant premium in initial capital costs in order to position the LEC to more easily upgrade to video later. This effectively amounts to cross-subsidization of video service provision with telephony service and is unfair both to consumers and to the cable companies. Ideally, the incremental cost of adding advanced interactive services later should be paid by those consumers requesting those services and not by telephone rate-payers. The current model indicates that the cost to consumers amounts to over $300.[9] If FTTC costs continue to drop, however, the argument could begin to shift, especially because FTTC provides much greater flexibility to handle unexpected future growth than does DLC. With FTTC, the fiber in the distribution plant can easily carry additional traffic, without the need to pull more cable except, perhaps, in the drop plant. With DLC, unexpected growth can require pulling new copper distribution plant.

This analysis also sheds light on the ability of an HFC system to meet the demand for Internet access. HFC systems cannot readily offer significant upstream bandwidth. This means that if, for example, a significant number of consumers want upstream bandwidth for outgoing video from their home-based Web servers, HFC would not be well-suited to handle their demands. The same goes for businesses that might want to be content-providers and not just content-consumers. From a public policy perspective, therefore, if the long-term goal is to empower everyone to be a producer as well as a consumer of content, then policy-makers should be opposed to HFC systems due to the asymmetric nature of the architecture.

The results of this analysis confirm that there are indeed economies of scope in the provision of interactive video and other broadband services. It is easy to see from Figure 7 that it costs significantly more to provide services via overlaid architecture pairs (e.g., HFC over FTTC) than it does via architectures that are upgraded by adding more equipment like electronics.

Summary and Conclusions

The results of this analysis clearly indicate that the All-HFC architecture is the most economical option for the provision of interactive video services, particularly at small penetration rates. However, because the all-FTTC option is relatively insensitive to penetration, one can expect it to rival the all-HFC system with increasing demand for IVS services. As time goes on, FTTC component price reductions should also narrow the gap between the FTTC-based systems and the all-HFC architecture. Moreover, the uncertainty in the all HFC scenario is probably larger on the upside: managing ingress noise could turn out to be costly over time. The analysis also shows that the cost of the all-HFC system is highly sensitive to bandwidth requirements, a fact that should make the FTTC-based systems more competitive in the long-run. The economics of the five scenarios considered also shows that cable companies have a very good chance to compete in the new arena of broadband interactive services. Indeed, the results show that at low penetration-rates, cable companies would be able to offer services at much lower costs than the telephone companies.

Our analysis also sheds light on what the FCC's Video Dialtone mandate costs the LECs and the rate-payers. Our results show that this requirement forces the LECs to spend an additional 100-300 dollars per home passed to build their networks. In addition, this mandate assumes that consumers would not mind paying more money to purchase the advanced services that the LECs would then be able to offer. If this assumption ends up being invalid, the telephone companies would be at a distinct disadvantage. Another important conclusion to be drawn is that the Cable Act's insistence on supporting cable-ready TV receivers currently does not represent a binding constraint. This is because the all-HFC system is currently the most economical system. However, the analysis shows that long-term demand increases, and price reductions in FTTC components could result in FTTC-based systems that could compete favorably with HFC-based architectures. Under these assumptions, the cable-ready rules, absent a waiver from the FCC, could effectively force an LEC to give up the option of Title VI regulation in order to be free to choose to build a pure FTTC system.

The results of our models also show that a baseline FTTC for voice has higher initial capital costs than DLC. This means that should the telephone companies build voice-only FTTC systems and put all the costs into their telephone service rate-base, consumers could end up paying more for telephone service, the difference being the added cost for positioning to provide video infrastructure. This would amount to cross-subsidization of video service provision with telephony service provision and would also be unfair to the cable companies. The counter-argument that is made by the LECs is that FTTC will have lower operating costs due to its enhanced flexibility to respond to requests for second and third lines per home. Lastly, the results of our analysis indicate that there are indeed economies of scope in interactive service provision and that, despite their low-cost, HFC systems are not well-suited to provide interactive services where consumers also produce content. In choosing the best technology at the cheapest cost, therefore, the FCC should bear in mind that the asymmetric nature of HFC might, in the long-term, render it unsuitable for such interactive services.

References

1. Nosa Omoigui, Marvin Sirbu, Charles Eldering, Nageen Himayat (in preparation): "The Economics of Competing Integrated Broadband Architectures"

2. Robert Mason, Nageen Himayat, Charles Eldering, Nosa Omoigui, Marvin Sirbu: "Overview of Hybrid Fiber-Coax and Fiber-in-the-Loop Architectures" (presented at the National Fiber Optic Engineers Conference, June 18-22