FTTH architectures face interconnect issues inside the final mile

Installing Fiber-optic Networks

As fiber extends the final few "yards" to the home, a great debate will determine the most efficient and cost-effective means of "last yard" connectivity.

George O'Mara and Scott McCrearySumitomo Electric Lightwave

Advances in passive-optical-network (PON) architectures and some proprietary passive optical solutions have been shown recently to reduce the initial costs of fiber-to-the-home (FTTH) systems. The result has expedited the cost-effective means to deploy voice, data, and video through fiber for the "last mile" to the home. Decisions and debate now focus on cost-effective solutions, the location within the home of the optical-network unit (ONU) or optical-to-electrical converter, and the means by which the last few "yards" of the network will be connected.

Major contenders in the FTTH race are the regional Bell operating companies, competitive local-exchange carriers, cable-television companies, and private real estate development firms. All are experimenting with various alternatives, primarily a plug-and-play configuration (pre-terminated connector-to-connector solution), an all-fusion spliced network, or a combination of both. The interconnect issue for the final few yards to the home concerns the alternative approaches with which to complete the system from the passive optical splitter to the optical-to-electrical converter (or set-top box) located in the end-users' premises.

Arriving at an "ideal" cost-effective solution depends on the advantages and disadvantages associated with the two major alternatives when considering ease of access into the network for troubleshooting and maintenance. Another criterion for making the most appropriate interconnect choice within various network designs is to consider the best approach to meeting future scalability requirements, which are driven by customer demand for ever-increasing bandwidth to facilitate the industry's promise to provide a broader range of better and faster Internet services.

The main elements of the PON access network are the optical line terminal (OLT); a passive splitter/coupler that directs wavelengths at the network interface point; drop and distribution cables; a network interface device (NID); and one or more ONUs. The NID is attached to the side of the home and considered an optional component in some network topologies. The OLT can be located at the communication company's office or at the headend node of the PON. The ONU can be positioned on street locations, near buildings, or on the user's premises. Depending on where the PON terminates, the system can be described as fiber-to-the-curb (FTTC), -building (FTTB), or -home.

Although a framework for close-in fiber-optic installations is being developed by an organized effort called the Full Service Access Network initiative, many proprietary topologies exist, including combination FTTH and FTTC designs and cable-TV-specific configurations that integrate existing coaxial cables. Regardless of the variety of con figurations proposed by fiber-to-the-x players, decisions about the various interconnect alternatives need to be made.

Figure 1. A typical all-fusion network's topology bypasses the network-interface device to bring individual drop cables directly to the home's optical-to-electrical converter. This approach extends flexibility to the installer for the most convenient placement of splice points.

Focusing the scope of discussion to FTTH, two of the more popular interconnect approaches-all-fusion and plug-and-play (connectorized)-being considered by major incumbent local-exchange carriers (ILECs) and other FTTH leaders warrant exploration. A summary of the advantages and pitfalls of these FTTH interconnect approaches with respect to reliability, cost considerations, and future scalability can assist in making appropriate interconnect decisions that can be applied to most other network topologies.

The access to the network in the all-fusion scenario is located at a mid-span splice point. This splice point can be an aerial splice closure or an outside-plant cabinet/closure used for buried cables and conduit installations in more congested city locations or where underground or conduit cable-plant networks are more common. Power supplies in pedestals and cabinets are not needed since the electronics are housed inside the home. Since aerial installations are less expensive and less disruptive to neighborhoods, most FTTH trials have been tested in areas where most of the houses have aerial connections.

For the all-fusion installation, the installer must first access the distribution cable at a mid-span location and cut as many fibers as needed for fusion splicing to the splitter/coupler. There would be additional fusion splicing operations of the tail fibers exiting the splitter, which would be spliced to individual drop cables brought directly to the home's optical-to-electrical converter, bypassing the NID (see Figure 1). The all-fusion approach clearly extends to the installer or contractor the flexibility of placing the splice point wherever it is most convenient.

For the pre-terminated connector-to-connector solution adopted by proponents of the plug-and-play approach, the access point to the fiber network from the distribution cable is at a pole-mounted or pedestal-mounted splitter terminal.

Working with the splitter terminal, the installer must first access the distribution cable at mid-span and cut the fibers that need to be connected to a pre-terminated jumper, which would be brought to the splitter terminal. The connection between the distribution cable and jumper would most likely require a splicing operation. An alternative operation would be to use customized distribution cable with preinstalled connectorized nodes, which can then be directed to a splitter pedestal or designated distribution point. This process would eliminate the initial need for fusion splicing, resulting in a true plug-and-play concept.

Figure 2. In a typical plug-and-play network, the drop cables exiting from the splitter terminal are plugged into the splitter output and directed to a network-interface device located outside the customer home.

The drop cables exiting from the splitter terminal in both options described above would be plugged into the splitter output and directed to the NID located on the side of the customer's home. A final terminated jumper cable would then go from the output on the NID to the optical-to-electrical converter in the home. The converter is also pre-terminated (see Figure 2). As mentioned earlier, advocates of the all-fusion approach bypass the NID altogether.

The connector-to-connector approach can, therefore, be plug-and-play from the splitter terminal to the home or from the distribution closure to the home. Although the term "plug-and-play" is often used to describe this interconnection, few operations in the industry are as automatic as the term implies. Like the all-fusion approach, the plug-and-play configuration brings to the installer advantages and challenges.

High attenuation losses through couplers and multiple path consolidation in today's network make fault locating a serious challenge for both the all-fusion and plug-and-play interconnect approaches. Fault detection by viewing the network from the headend or CO through a splitter is quite difficult with the use of Rayleigh backscatter optical time-domain reflectometers (OTDRs). The distances between events such as splices, couplers, and connectors are greatly shortened and beg the need for high-resolution measurement devices like a Fresnel-mode OTDR.

Clearly, the greatest disadvantage of the all-fusion interconnect method is its reduced access points for testing, given its direct path from headend/

CO to the end user's ONU/converter. The network configuration of the all-fusion approach requires that troubleshooting be done inside the premises, adding a potential inconvenience to the end user.

A partial WDM solution is to introduce into the currently adopted configuration an external NID with a coupler or to place a WDM coupler in the home's optical-to-electrical converter, allowing the installer/technician access to the network. The technician can either use a Fresnel-mode OTDR that can test through the coupler or a standard Rayleigh OTDR that will test up to the coupler point. With standard testing devices, the technician would be required to shut down that link so the testing equipment would not affect the performance of the transmitter and vice versa. The technician would then make measurements with either a power meter or OTDR.

This measurement technique, however, requires additional couplers and filters to the network path for technician access and the achievement of clear readings. In addition, manufacturers would need to develop power meters with wavelengths other than 1,310 or 1,550 nm to accommodate the added channels in the link.

The advantage of the many access test points associated with the plug-and-play configuration is nullified by the use of splitters/couplers within the network path. The use of a Fresnel-mode OTDR offers the only visual testing solution for fault decibel-loss detection, since it is capable of seeing through couplers. However, the Fresnel-mode-capable OTDRs are more expensive than the industry-standard testing equipment and must be added into the cost equation. Moreover, the network links would need to be well-documented so that a technician could interpret images such as Fresnel reflections and ghosting displayed by the unit. The thorough documentation required for successful troubleshooting and maintenance also adds to the overall cost of the installation.

Overall reliability issues, such as the ease of network access for testing, need to be factored in with each approach's contribution to loss budgets. The all-fusion approach undoubtedly generates lower link losses than the plug-and play, given that the average attenuation loss for SC, FC, and ST-ultra-physical-contact (UPC) connectors is 0.35-dB versus an average 0.05-dB loss from a fusion splice. For applications in which loss budgets are a major concern, such as the transmission of broadcast analog video through a traditional analog cable-TV system, the fusion-splicing option will offer significantly lower end-to-end attenuation loss. Moreover, connectors in the outside plant are more susceptible to the environmental fluctuations that lead to inconsistent attenuation performance, thereby driving the necessity to build more loss-budget tolerance into the network design-yet another consideration when judging overall reliability.

The primary advantage of the all-fusion approach is its ease of scalability, given ever-increasing next-generation demands for high bandwidth. As network protocols become more advanced allowing for faster and faster speeds (multigigabit demands in one to three years), the more attractive becomes the all-passive, all-fusion configuration that provides a direct link to the home. Unlike the plug-and-play approach, with connector type limitations for the amount of bandwidth accommodated, the fusion operation can accommodate 155 Mbits/sec to 100 Gbits/sec or more.

The right interconnect approach is crucial in FTTH applications. Fusion splices will be necessary regardless of the approach, however.

To the degree by which the all-fusion approach's greatest advantage is its ability to maximize scalability to meet the growing demand for increased bandwidth, it is also the greatest disadvantage for the plug-and-play configuration. The plug-and play's connector-to-connector solution is susceptible to the creation of backreflections, which cause pulse distortion and phase noise affecting high-speed transmission. The backreflection is a result of both Rayleigh backscattering and Fresnel reflection, which could contribute to high bit-error rates. Networks with speeds of up to 200 Mbits/sec can accommodate interconnection reflections with standard PC connectors. Gigabit speeds require the use of low backreflection connectors.

Ceramic ferrules-FC/PC, ST/PC D4/PC, and SC/PC-produce terminations of -32-dB backreflection with hand-polishing. However, with proper automatic machine polishing, levels of backreflection of -55 dB are achievable in UPC connectors, given sizable investments in equipment. This operation can be performed in the field but with difficulty and added time considerations. For many high-bandwidth requirements, such as analog cable-TV transmission above 600 MHz, for example, -60 dB or above is necessary.

To accommodate growing bandwidth needs, the plug-and-play configuration must invest in the use of angled-physical-contact (APC) connectors with backreflection levels of -60 dB or beyond-a theoretically practical but costly solution. APC connectors cost about double that of standard SC/UPC connectors but exhibit no Fresnel reflection if polished and mounted correctly. There are, however, tradeoffs between the use of APC and PC connectors. Although APC connectors have lower backreflection, for example, they also generate higher insertion losses.

At first glance, it is easy to assume that the plug-and-play configuration saves on initial labor costs associated with the all-fusion approach. However, the plug-and-play requires significant labor time to perfectly measure the pre-connectorized lengths of distribution and drop cables. If the lengths are not accurate, the excess cable will have to be stored or the pre-connectorized ends cut and re-terminated, thereby adding to installation costs.

In combination FTTH and FTTC designs, the addition of new customers would require new cables pulled, often along the same path as previous drops. However, in areas where both current and future network requirements are known, the plug-and-play deployment becomes increasingly more attractive. The plug-and-play, however, must account for upfront engineering costs and the risks of excess inventory of pre-engineered standard lengths. For material costs, it is imperative that there is compatibility between connector and connector, often requiring that the same manufacturer be used to control tolerance variations across the connectors. This requirement can lead to price inflexibility in long-term deployments.

Because of the many unique FTTH topologies being tested within various PON networks, installation costs for both the all-fusion and plug-and-play can be quite variable. Some topologies, for instance, have opted to incorporate specialized distribution and drop cables. Several companies, including Sumitomo, have developed innovative cable designs, for example, that lower fusion-splicing time and generate overall savings in installation costs. The use of these cables could offset the labor and deployment cost differences between the all-fusion and plug-and-play interconnect approaches.

Cost variations in overall installation comparisons also depend on whether the deploying company has preexisting inventories of fusion-splicing equipment and skilled labor, which can offset the upfront capital required for deploying the all-fusion operation.

The facilitation of scalability offered by the fusion approach must be weighed against its greatest disadvantage: the difficulty in accessing the network for testing and troubleshooting faults and high attenuation readings. In turn, the pre-terminated connector-to-connector solution must be weighed in terms of its shortcomings for accommodating growing bandwidth requirements and the associated costs. All the respective strengths and weaknesses in the installation process of both methods need to be considered, as well.

Choosing between the all-fusion, plug-and-play, or an incorporation of both depends on how each interconnect method can satisfy the requirements of a network's unique topology. By weighing the pros and cons of interconnect options, the right choice becomes less complex as the FTTH race continues the final few yards.

George O'Mara is the interconnect products manager at Sumitomo Electric Lightwave's Communications Network Group (Research Triangle Park, NC), and Scott McCreary is an applications engineer and the team leader for Sumitomo's fiber-to-the-home product-development projects.