Fiber to the Home

VDSL and FTTC bring fiber to within 200 meters of the home. These are viewed by some as transition services on the way to bringing fiber from the carrier to inside the customer premises. Fiber to the Home (FTTH) is fast, is immune to electromagnetic interference, and doesn't rust, so it has lower maintenance costs for the carrier. Bell South is investigating the use of FTTH (in the Integrated Fiber in the Loop [IFITL] project), and similar studies are being conducted by the Canarie Project in Canada, NTT in Japan, France Telecom, and British Telecom.

Until recently, there have been a number of roadblocks to extending fiber all the way to the home, including installation costs, electrical powering, and the precision with which fiber must be handled. Recent technical innovations and application development indicate that each of these hurdles is being confronted, so the prospect of FTTH is not quite as farfetched as once imagined.

FTTH comes in two forms. First is a shared form very much like HFC. It has a shared forward channel and point-to-point reverse channels with a MAC protocol to arbitrate return traffic contention. This form of FTTH is called passive optical networks (PON). The second form is a point-to-point optical service wherein each residence has a dedicated optical channel to the carrier that is shared only by the devices in the home, called dedicated FTTH.

Passive Optical Networks

Passive Optical Networks (PON) is being standardized by ITU SG-15 as Recommendation G.983, "High Speed Optical Access Systems Based on Passive Optical Network (PON) Techniques." The Recommendation describes two service options:

• Symmetric service of 155.520 Mbps (OC-3)

• Asymmetric service of upstream 155.520 Mbps (OC-3) and downstream 622.080 Mbps (OC-12)

The transmission medium consists of one or two single-mode fibers in accordance with ITU-T Recommendation G.652. Bidirectional transmission is accomplished by use of either a wavelength division multiplexing (WDM) technique on a single fiber, or two unidirectional fibers. For cost reasons, the one-fiber system is preferred. In the one-fiber system, the forward path shall be transmitted at 1480 to 1580 nm. The operating wavelength range for the upstream direction shall be 1260 to 1360 nm. The recommended characteristics of the fiber are shown in Table 5-2.

Table 5-2. Physical Media-Dependent Fiber Characteristics

Characteristics	Value
Fiber type	1.3 μm zero-dispersion fiber (ITU-T G.652)
Optical path loss	10 to 25 dB (ITU-T G.982 Class B) 15 to 30 dB (ITU-T G.982 Class C)
Path loss	15 dB
Maximum logical reach	20 km
Minimum supported split ratio	16- to 32-way split

Line coding is On/Off keying. Light ON means a binary 1, and light OFF means a binary 0. This simple form of line coding reduces the cost of the optical components and is highly robust.

FTTH PON looks a lot like hybrid fiber coaxial cable networks (HFC), as opposed to the Very High Bit Rate Digital Subscriber Loop (VDSL) architecture, with which it is often compared. Simply replace the HFC drop wire with single-mode fiber, and reduce the number of homes per cluster from 500 to 1. The nomenclature is different, but the startup functions of FTTH are familiar to observers who know HFC. Some similar functions are ranging, the requirement for a MAC protocol to arbitrate upstream bandwidth utilization, control of transmitted power from the residence to the carrier facility, and a broadcast-based, plug-and-play registration process. In addition, all the Internet startup functions are required, such as the assignment of IP addresses and packet filters.

A schematic of a PON system is shown in Figure 5-2.

The key elements we discuss are the Optical Line Terminator (OLT), Optical Distribution Network (ODN), and the Optical Network Unit (ONU).

The basic architectural elements are similar to other high-speed access networks. The OLT is the equivalent of the DSL Access Multiplexer (DSLAM) of xDSL networks, or the Cable Modem Terminal Server (CMTS) of HFC data networks. The OLT couples the transport network, usually an ATM backbone, to the local loop network and terminates and provides much of the control for the distribution network. The ODN is the local loop distribution network, and the ONU is the in-home device that connects the home network to the distribution network.

Optical Line Terminator

The OLT would reside in a telco central office or in a loop carrier system. Its functions include these:

• Optical transmission and reception

• Control functions in the distribution network, such as control of transmitted power from the residence to the carrier facility, forward error correction, and interleaving

• Enforcement of the MAC protocol for upstream bandwidth arbitration

• Coupling of the distribution network with the ATM transport network

• (Optional) Enforcement of higher-level protocols, such as address resolution, address assignment, tunneling, and conditional access control

• (Optional) Switching or cross-connection, to relieve the transport network of switching responsibilities

Optical Distribution Network

Within the larger PON architecture, the optical distribution network is comprised of single-mode optical fibers and the passive optical components, primarily optical splitters—hence the word passive in Passive Optical Networks. The Optical Distribution Network (ODN) offers one or more optical paths between one OLT and one or more ONUs.

Like electronics, optics suffer from attenuation and are measured in dB. The ITU specifies that the variation in path loss from one home to another be bounded. That is, one home may be closer to the OLT than another home connected to the same OLT; the standard limits the difference in path loss between the two homes to 15 dB.

The distance between the OLT and ONU can be as long as 20 km.

Optical Network Unit/Residential Gateway

The ONU provides the necessary functionality to connect the carrier fiber to the residence. Whereas the ONU for HFC networks serves 500 to 2000 homes, the ONU for FTTH serves one home. It performs functions similar to a digital TV set top box, an HFC cable modem, or an ADSL ATU-R. Its functions include these:

• Optical transmission and reception.

• Cooperation with the OLT to control transmitted power from the residence to the carrier facility.

• Forward error correction and interleaving.

• Enforcement of the MAC protocol for upstream bandwidth arbitration, in cooperation with the OLT.

• Coupling of the distribution network with the in-home network medium, such as plastic optical fiber (PoF), IEEE 1394 Firewire, or some form of metallic wiring. Coupling with an in-home network will involve some form of protocol handling as well, such as speed matching, buffering, and framing

• ATM multiplexing. Multiple devices in the home will be connected to a single ATM port from the home to the carrier. The ONU and residential gateway are responsible for multiplexing the sessions on the fiber link.

Because of the complicated nature of the interface functions, a component called the residential gateway or home network gateway will possibly decouple the purely optical functions of an ONU from high-layer protocol functions.

Some believe that an outdoor ONU would increase location flexibility, facilitate maintenance access, and provide a cleaner demarcation for the network interface. Others believe that the FTTH system will benefit more from the cost reduction of an indoor ONU, thereby making it look more like a piece of consumer electronics. As of this writing, these questions and other specifics of the ONU are reserved for further study and product development.

Proprietary variations of PON exist. For example, Alcatel has developed a product line called SuperPON. It distributes 2.4 Gbps in the forward and shares 311 Mbps on the return from up to 2048 clients. The range is up to 100 km.

Principles of Operation

For downstream data, ATM cells from the Internet, corporate network, or some other source traffic are transmitted through the Carrier ATM Transport Network and received by the OLT. Traffic is forwarded to the ODN, which consists of single-mode fiber that connects to an optical splitter. The job of the splitter is to take the downstream optical wavelength and replicate it optically so that each ONU receives the same transmission. In this way, one laser transmitter at the OLT serves multiple residences. Splitters replicate on the order of 16 to 32 tributaries.

For upstream data, the ONU in the home obtains data from television set tops (for example, channel selection) or from personal computers in the home. The ONU requests permission of the OLT to transmit upstream. The OLT processes requests for bandwidth and grants credits to the ONU for a specific number of ATM cells to be sent by the ONU. The OLT and the ONU must cooperate to arbitrate congestion on the return path. The process of media access control uses a credit-based scheme in which the carrier issues credits for time slots into which the client can send return path data.

The use of time slots means that all home transmitters connected to a common OLT use a common upstream frequency. Therefore, a single-laser receiver at the central office can receive traffic from all residences. This is an important cost-saver and is why shared FTTH is seen as more practical than point-to-point or dedicated FTTH. On the other hand, if the cost of laser components declines, it is possible to consider the case of dedicated fiber connections to the home.

All current PON proposals assume the use of ATM cells to the residence. The application may or may not be an ATM device; if not, the residential gateway would be responsible for the adaptation of the legacy protocol such as IP or MPEG into and out of ATM. This process is called segmentation and reassembly (SAR) and is a common function for ATM devices.

The downstream signal is broadcast to all ONUs on the PON. Each upstream transmission from each ONU is controlled by OLT. A single residence can receive a variety of services, such as data retrieval, television reception, and telephony. Each of these cells is transmitted using TDM techniques. The ONU demultiplexes the optical line format into the various services.

Broadcast television will be supported using switched digital video techniques, just as in the FTTC case. Even though the return path speed of FTTH is faster, the speed at which tuning occurs is still an issue because the critical factor is how fast the access node can create the virtual channel from the content provider to the subscriber.

PON Issues

TDMA upstream optical reception and transmission causes three main difficulties that do not arise in point-to-point transmission:

• A packet arriving from an ONU close to the OLT may have a signal level much higher (due to lower path length-dependent losses) than a following packet arriving from a distant ONU, and the receiver must adapt to the new level within the guard time between packets.

• The timing of the upstream signals is complicated by a path delay that differs from ONU to ONU. In addition, the time-dependent variations in path length—for example, as a result of temperature changes—can also be very different from ONU to ONU. The timing of transmissions from each ONU must be adjusted so that the packets arrive at the OLT separated by the guard time.

• The sum of the spontaneous light from several tens of (nontransmitting) lasers held at threshold is comparable to the signal level from the one transmitting laser, leading to a significant increase in the bit error rate in PONs. This effect is accentuated during the reception of packets from distant ONUs.

In part because of these concerns, future consideration may be given to a dedicated FTTH topology.

Dedicated FTTH

PON is a shared topology; up to 32 residences can share 622 Mbps of bandwidth. If all residences are active, average bandwidth per residence will be less than HFC. This would seem to be disappointing, given the allure and reputation of FTTH.

As more residences subscribe to FTTH, the cluster size will drop from 32 down to as few as 8 because service providers will have proper funding to penetrate the splitters farther into the neighborhood. Ultimately, each home may have a dedicated fiber circuit to the carrier.

New development is studying ways to use Dense Wavelength Division Multiplexing (DWDM) to create private virtual paths into the home. DWDM-based systems will enable carriers to allocate wavelengths to customers, providing dedicated bandwidth of up to 155 Mbps in both directions to each home. Figure 5-3 illustrates the use of DWDM in a point-to-point FTTH environment.

Figure 5-3 depicts content from the Carrier ATM Transport Network that is received at an ATM switching function. In the forward direction, content going to Jimmy is sent on one frequency, and content going to Rosie is sent on another. These frequencies are multiplexed onto a single-fiber loop by a DWDM. At the other end of the loop, the frequencies are demultiplexed and forwarded to the respective households.

A DWDM is basically a frequency division multiplexer for optics. Different frequencies, or colors, travel simultaneously on a single strand of glass fiber. The subscriber will be allocated an upstream and downstream frequency for his exclusive use. In the scenario illustrated in Figure 5-3, Jimmy receives on frequency λ₁ and transmits on frequency λ₂; Rosie receives on frequency λ₃ and transmits on λ₄.

Current versions of DWDMs multiplex 32 or more frequencies per fiber. Certainly, this number will increase with the introduction of lower-cost laser transceivers. For now, however, DWDM is used exclusively in carrier backbones because of cost. Each frequency requires a separate laser. To have a point-to-point FTTH solution, there needs to be further cost reduction with multifrequency lasers.

The advantage of point-to-point over PON is that there is no need for a contention-resolution protocol, and there is a greater degree of privacy. The drawback is cost. With point-to-point, a laser transceiver must be present at the head end for each subscriber. Furthermore, because of possible changes in subscriber usage, it will be necessary to remotely control the transmit and receive frequencies of the laser transmitter in the home.

Secondly, instead of having a passive optical splitter and multiplexer in the distribution network, there is a requirement for an optical switch. An intelligent, active switch is necessary to make sure that a given frequency goes to the correct subscriber.

Therefore, enhancements in lasers and optical switching are required before serious consideration can be given to point-to-point FTTH. But, in the distant future, this will be given further consideration.

Ethernet Local Loop

Other methods of connecting the OLT with the ONU are under consideration. The Palo Alto FiberNet ( Palo Alto FiberNet ) is a group of citizens in Palo Alto, California working with Palo Alto Utilities ( Palo Alto Utilities ) to introduce FTTH to homes and schools. The group's approach uses gigabit Ethernet switches to move IP packets over 10/100 Ethernet to homes. No ATM, and no telco, yet. Just rights of way, dark fiber, Ethernet switches, and a very technical and well-heeled citizenry.

Another feasibility test is underway in Stockholm, Sweden, conducted by Ericsson, which uses a packet-over-fiber infrastructure. A full-duplex Ethernet local loop is well-developed technology that can support neighborhoods perhaps in the hundreds. After all, switched Ethernet scales reasonably well for business environments. Ethernet can carry MPEG frames for video and is obviously well-matched for data transmission.

The issue with Ethernet is quality of service (QoS). But given the relatively small number of trial participants and the quantity of bandwidth in the core of carrier networks, Ethernet can be a quick and dirty implementation for no other reason than to validate that a consumer market exists for carrier-based Fast Ethernet service.

Benefits of FTTH

FTTH is viewed as the end game for access networks for a variety of reasons. The benefit to the user is greater speed; the benefit to the carrier is lower operational and provisioning costs. By "end game" we mean that the ONU is steadily approaching the home. When it gets there, the greatest possible performance can be achieved.

Return Path Bandwidth

Residences will benefit from more bandwidth—especially in the upstream direction, compared to HFC, xDSL, or FTTC—and thus it is the most appropriate technology for residences to host their own Web sites. Businesses can take advantage of high-speed return for corporate Web hosting.

Remote Provisioning

A key benefit of FTTH for the carrier is its capacity for remote provisioning. Remote provisioning refers to the capability of the carrier to configure customer service from the central office without incurring the cost of a truck roll. Estimates for the operational costs of sending a field engineer to the home vary widely but are bounded by figures of $200 to $500 per incident.

Using remote provisioning, a consumer can elect a service-level agreement on Day 1 that includes 10 Mbps service. Later, the same customer might want 20 Mbps service and, later still, as little as 2 Mbps. Because the bandwidth to the consumer is greater than any of these options, the carrier need only impose some flow-control measures at the central office to restrict consumer bit rate to the requested service level. Eventually, given improved network-management techniques, consumers could be able to conduct provisioning for themselves.

The carrier or the consumer can configure bandwidth for exclusive use for relatively long periods of time (days, or even hours). Remote provisioning would be subject to available link capacity and pricing, but many of the operations and maintenance costs could be reduced.

Remote provisioning of bit rate is enabled by the use of ATM to the home. ATM has the capability to support usage requests expressed in different semantics. Such semantics include peak rate bandwidth, average bandwidth per unit time, and the variability of transmission delay. Of course, monetary costs would be associated with such flexibility, but this would offer greater control for the user and lower provisioning cost to the carrier.

Distribution Network Cost Reductions

In addition to remote provisioning, another cost benefit for the service provider is reduced maintenance costs in the distribution network. These reductions come from reduced powering costs and the reduced costs of outside electronics and cabling. The cost of power to the ONU will be passed to the customer, unlike today's phone service, where the electric power to the telephone is supplied by the phone company. Although this is a small amount of power, the total power consumed by millions of phones adds up.

Greater Reliability

Optical components are immune to RF interference, such as ingress and crosstalk. They don't rust, which extends the usable life of components in the field. Optical signals take longer to attenuate than electrical signals through metal, and there is less corrosion. This all suggests that optical signal quality is superior to that of wired infrastructures and will likely remain so as the infrastructure ages. Fiber also has fewer active repeaters than metallic networks for equivalent distance. This contributes to greater reliability.

Full-Service Networking

FTTH and HFC are alone in their capability to provide full-service networking (FSN). FSN refers to the capability to provide high-bandwidth, multichannel services, such as broadcast television, and narrowband services, such as telephony. Marketers dub this feature futureproofing, which implies the capability to accommodate whatever services and content the future holds without changing the infrastructure. Over time, therefore, life-cycle costs are minimized.

Challenges of FTTH

Significant progress has been made in defining requirements and an architecture for PONs. But major problems remain, which promises to delay the availability of standardized FTTH for years to come. Despite the issues, FTTH will continue to be considered an important goal, primarily by the world's telephone operators.

CPE Costs

Lasers are expensive, and the consumer electronics industry has not had the time or the motivation to incorporate lasers into consumer products. Furthermore, there are cost issues of home wiring and residential gateways to support FTTH.

Carrier Capital Costs

Reduced electric bills, reduced provisioning costs, and reduced costs of outside electronics and cabling are long-term benefits for carriers.

But for the short term, major costs are associated with the startup of service. Among the key costs are installing the fiber (which may involve digging); connecting the subscriber equipment in the home (which involves labor and precision equipment); capital costs of the OLT, splitters, and ONU; capital costs of laser transmitters and receivers; new maintenance; and test tools and training.

Central office equipment costs have decreased because reductions in the costs of lasers and wavelength division multiplexers have been considerable in recent years. However, more cost reduction is required to make subscriber equipment FTTH costs competitive with wired infrastructures. For the moment, the startup costs outweigh the operational cost savings and might continue to do so for years to come.

Standardization

While there is basic agreement on ITU SG-15 G.983 standardization, many key elements of an end-to-end architecture have yet to be decided. Among these elements are powering, encryption and conditional access, zapping protocols, startup procedures, and inside wiring. Solutions today are vendor-specific. The lack of standardization means that carriers are reluctant to commit to large-scale commitments.

Power to the ONU

Telephones receive electrical power over the copper loops that carry the voice. Because fiber does not conduct electricity, powering of remote electrical elements in FTTH systems is one of the major issues associated with the deployment of FTTH. A requirement for widespread FTTH is agreement among regulatory agencies, telcos, and consumers regarding power to the ONU. Two powering options exist—namely, network powering and subscriber powering. A mixed architecture, referred to as hybrid powering, is also possible.

Network powering uses an overlay network of metallic wires in which a number of ONUs receive power from the carrier network. It is anticipated that legacy copper wire will be repurposed from carrying voice to carrying power. Even so, network powering for the ONU is costly for the carrier. For FTTH systems to be economically practical, the ONU needs to be powered from the consumer's outlets, with a battery backup provided at either the consumer or carrier's expense in case of emergencies.

With subscriber powering, all subscribers are responsible for the power feeding and therefore pay the electric bill. Some operators expect that the customer may become responsible for the proper operation of the battery backup, as with cellular or cordless telephones today. Subscriber powering is the lowest-cost option for the carrier.

Arguments, primarily from the carriers, will assert that the residential powering problem is diminishing. The use of cellular phones is considerable, and they can be used in an emergency. Other alternatives are batteries, which can operate phones for 8 to 24 hours. Whether this can pass muster with regulatory agencies or the public at large is to be determined.

Switched Digital Video

The PON specification has insufficient bandwidth to offer full-broadcast television. Direct-to-home digital satellites offer up to 144 channels of broadcast television. Upgraded HFC systems will be capable of 700 MHz of analog and digital television. Such systems can provide 50 channels of analog TV and up to 1.5 Gb of digital TV. This is far more channel capacity than an FTTH system with 155 Mbps or even 622 Mbps of shared bandwidth. Therefore, a developing technology called switched digital video (SDV) is indicated to offer a competitive service supporting broadcast TV and VoD. The elements of SDV will be discussed in Chapter 8, "Evolving to RBB: Systems Issues, Approaches, and Prognoses."

Fiber-Handling Problems

Splicing, handling, and bending of fiber are of particular concern. Fiber is a difficult medium to manipulate. When splicing two segments of fiber optic cable end-to-end, the part that actually transmits light is only 50 microns (millionths of a meter) in diameter. If the fiber segments are not perfectly aligned, optical impairment results. Splicing fiber cable requires expensive precision equipment. Technical innovations in development will increase the tolerances with which fiber can be handled, but these need further refinement. Bending is also a consideration—fiber cable cannot be bent excessively without incurring severe attenuation.

Two broad classes of splices exist: fusion and mechanical. Fusion splices operate by melting the adjoining segments together. They provide very low losses at the splice point (0.1dB). Multifiber splicer equipments are available that allow splicing of several fibers in a single operation, thereby saving labor. This is important, because splicing costs in the field are sensitive to labor cost and splice amortization. Costs range from $10 to $40 per splice, depending on these two factors.

Mechanical splicing is based on direct alignment into a simplified connector. Mechanical splices are less expensive than fusion splicing ($10 to $15 per splice), but they create high optical loss and reflectivity. Because of the requirement for cleaner signals, the trend is toward the use of fusion splicing, despite its cost.

In general, electrons traveling through metal wire are much more forgiving than photons traveling through fiber. The increased precision and costs associated with fiber installation as compared to metal wire are ongoing challenges to FTTH.

Connectors

Related to fiber-handling problems is the problem of optical connectors. Optical connectors are used to insert fiber into various active and passive pieces of equipment, such as splitters, OLTs, and ONUs. Connectors are a source of signal loss, but technology has advanced to the point where there is a low insertion loss of around 0.2dB per connection. Techniques have been successfully implemented regarding cutting, polishing, physical contact, and refractive index matching.

Many connectors have ceramic packaging, or ferrule, which accounts for a significant fraction of the total cost of the connector parts (30 to 50 percent), so cheaper materials such as glass or plastic are now under development.

Despite advances in splicing and connectorization, field handling of fiber requires well-trained technicians and expensive tools.

Encryption and Wiretapping

Because the cable is shared, data will be encrypted. Encryption, and the fact that fiber is more difficult to wiretap than metallic cables, means that some method of enforcing court-ordered wiretaps must be implemented before widespread use of FTTH will be allowed in the United States and other countries worldwide.

Already, conflict has begun between law enforcement and the telecommunications industry over wiretapping of digital telephony and voice over IP. Techniques used to wiretap analog voice calls do not work when voice becomes a digital application. Law enforcement is asking that modifications be made to digital telephone equipment to facilitate the same level of wiretapping that is possible with the current analog voice network. These modifications involve some costs, resulting in objections from telecommunications carriers and Internet libertarians. This issue is a harbinger of controversies to come, when all transmissions to and from the home will be digital.

Market Acceptance

So far, the interest in FTTH is mainly push from carriers and their equipment suppliers. Very few content providers and consumers are demanding FTTH service. The roster of participants in FSAN are all carriers, mainly European, and their providers. So far, the volume of effort behind FTTH significantly lags behind the standardization and product development work enjoyed by HFC and xDSL. More basically, it is yet to be determined what levels of pricing the public will pay for high-speed services to the home, especially services that do not provide for efficient delivery of broadcast television.

Transition Issues

The success of xDSL and HFC could defer the development of FTTH. Although the prospect of sharing up to 622 Mbps among a handful of homes is enticing, the fact remains that PON offers less bandwidth in the forward direction than HFC. Furthermore, HFC is operational in millions of homes, whereas FTTH is only in the trial stage. Therefore, FTTH service providers will not have the breadth of programming that DBS and HFC have. It is likely that FTTH will have an advantage for pull-mode services, either VoD or pull-mode data, but it remains to be seen if push-mode or pull-mode services will provide the most lucrative funding model for RBB.