Fiber-in-the-loop topologies ease migration into full-service networks
The combination of Sonet ring protection and active star topologies results in a reliable and cost-effective fiber-in-the-loop platform
Bill McDonald and Paul Langner
Fujitsu network transmission systems inc.
By using a survivable feeder architecture, the remote deployment of fiber-in-the-loop, or FITL, host digital terminals in the customer serving area ensures the level of durability that emerging broadband networks need. In addition, the fiber-in-the-loop active star distribution topology exhibits low cost, flexibility, high bandwidth and consistency with accepted network practices. Moreover, combining fiber-ring protection and active star topologies into a cohesive loop-access network provides seamless migration into a full-service network.
Fiber-in-the-loop technologies are expected to be omnipresent during the next 12 to 24 months. Hybrid fiber/coaxial-cable architectures are proving to be limited and expensive for delivering high levels of telephony, digital and broadband services. As local exchange carriers prepare to install fiber-to-the-curb, or FTTC, architectures for broadband access, several deployment strategies must be evaluated, such as:
Locating the host digital terminal in the central office or at a remote terminal site
Using an active star or a passive optical network distribution topology between the host digital terminal and the optical network unit
Evaluating survivability and growth requirements
Exploring the migration path for upgrading fiber-in-the-loop networks to full-service networks.
To deploy fiber-in-the-loop networks, these architectural choices must be investigated relative to cost, effort and reliability to structure a full-service network. Equipped with network modeling tools and a valid migration strategy, network planners can accomplish this goal.
Full-service network migration
To meet customers` demands for more capabilities from their telecommunications service providers, network planners must determine the proper network architecture that satisfies present and future needs. Whether meeting existing services (for instance, traditional telephony, data and video services), expected services (such as broadcast video, video-on-demand and interactive multimedia television) and future services (such as high-definition television, video telephony/conferencing and home/office retailing), installed networks must be capable of providing these services immediately--in a few months or in a few years. Based on these requirements, a fiber-in-the-loop architecture represents the optimum choice.
The first step for network planners is to determine the best location for the host digital terminal. The obvious options are either in the central office or at a remote terminal site. A major obstacle in choosing the central office concerns the effects of a catastrophic failure. For example, in 1988, a fire destroyed a local exchange carrier`s central office in Hinsdale, IL. This service disaster disrupted almost 65% of the voice traffic and 85% of the data traffic in the southwest Chicago area. In fact, local and intra-local access and transport area services to some customers were not reinstated for a month.
Consequently, consolidating a large amount of network equipment, such as host digital terminals, in a central office risks complete service disruption because of a single failure. And when multiple carrier-serving areas are supported by a single feeder cable from the central office, all services are vulnerable to interruption because of a cable cut. Because an entire central office serving area could be devastated by a single failure, this option is not recommended for housing host digital terminals.
The other option is to place these terminals at remote terminal sites that are connected to a self-healing synchronous optical network, or Sonet, fiber-ring topology that, in turn, passes through at least two central offices. With a host digital terminal deployed in each carrier-serving area, only 1200 customers within that area would be affected by a single failure. With multiple host digital terminals supported in a single self-healing fiber-ring that is connected to two central offices, both fiber-cost savings and increased reliability are realized. With this option, a single central office failure or a single feeder cable cut would cause minimal service interruption.
The survivability of the Sonet fiber-ring becomes important to users as full-service networks are implemented. As carriers offer more services, consumers are expected to tolerate fewer interruptions. The remote terminal site option for host digital terminals is deemed the best choice for meeting present and future network demands.
Topology alternatives
Based on the conclusion that the best place for host digital terminals is in the outside plant portion of the network, the next step for network planners is to focus on the portion of the network between the terminal and the optical network unit located near a group of customers. The two prevailing fiber-to-the-curb topologies for configuring this network portion are the passive optical network and the active star topology.
In the early stages of fiber-in-the-loop evolution, the passive optical network implied a passive medium connected between the central office and the customer interface unit. The required electronics were emplaced in the central office and in the customer interface unit. Passive optical splitters were used to dispense the signals from the central office throughout the carrier-serving area. At that time, the telecommunications industry generally believed that minimizing the number of active elements in the outside plant facilities would reduce maintenance costs. However, this approach was eventually determined as impractical because of economic and lifeline powering considerations.
The passive optical network is still used, but it now identifies a hybrid system that uses a passive optical splitter between the host digital terminal and multiple optical network units that contain active hardware. The use of a passive optical network is an attempt to save fiber count and number of optical units used. Although it has these savings advantages, this network suffers from technology limitations, such as a lack of temperature-hardened optics for rates above 622 megabits per second, which restricts its ability to support bandwidth upgrades economically.
The other distribution system topology is known as active star, which is characterized by the use of dedicated optical transmitters and receivers and fiber pairs from the host digital terminal connected directly to each optical network unit. Optical splitters are not required. Because the transmission rate between the host digital terminal and each optical network unit in the active star is less than that of the passive optical network--155 versus 622 Mbits/sec, respectively--future upgrades to higher rates are attainable using existing technology.
Cost analysis
To assist network planners in analyzing costs, a distribution network model is assumed using accepted telephone company deployment guidelines. To avoid the single point of failure ramifications, the host digital terminals are located remotely for both topologies to ensure survivability.
In this network model, a host digital terminal is deployed at a remote terminal site and supports a 155-Mbit/sec bandwidth. The terminal is connected to four distribution areas, each through a serving-area interface. In turn, each of the four interfaces serves 20 optical network units. In sequence, each optical network unit supports 16 customers. As a result, 320 homes per distribution area or 1280 homes per carrier-serving area are linked to the host digital terminal.
The fiber-optic cable distance from the host digital terminal to a serving-area interface in each distribution area, known as a sub-feeder, is typically a maximum of 12,000 feet in length. However, to compare more completely the cost versus cable length of the sub-feeder for both topologies, the cable length is varied to a maximum length of 45,000 feet. Costs for splices and splice hardware for every 3000 feet of cable, as well as construction and installation costs, are included.
Passive network definition
The passive optical network architecture comprises the following parts:
Five 622-Mbit/sec long-reach optical transceiver modules at the host digital terminal. Long-reach optics are required to span the distance to 45,000 feet and to compensate for the 7-decibel insertion loss introduced by the optical splitters.
A 12-fiber management area at the host digital terminal equipped with associated jumpers, splicing, interconnect and storage hardware
A 12-fiber, armored, buried, sub-feeder cable with 10 active fibers
A serving-area interface cabinet equipped with 10 1:4 splitters with splicing, interconnect and storage hardware. Because temperature-hardened units are not available for optical rates above 622 Mbits/sec, the passive optical network host digital terminals located in remote terminals are limited to this rate. The splitters distribute the optical signal to four optical network units, each of which has access to 155 Mbits/sec of bandwidth.
A 622-Mbit/sec optical transmitter/ receiver module at each optical network unit.
The 622-Mbit/sec units at the host digital terminal are connected to the serving-area interface via 10 optical fibers. At the interface, each of the five transmit fibers is connected to a 1:4 optical splitter. Each of the four outputs from each of the five splitters goes to an optical network unit via fiber. The unit thus processes the 622-Mbit/sec signal and extracts the 155-Mbit/sec bandwidth intended for it. This signal is further processed for distribution of individual signals to each of 16 homes.
In the reverse direction, back toward the host digital terminal, signals sent from the homes are combined into a 155-Mbit/sec signal in the optical network unit, which then inserts 155 Mbits/sec of data into the 622-Mbit/sec signal for upstream transmission. Back at the serving-area interface, the 622-Mbit/sec signals from four optical network units are combined using a 1:4 splitter that works as a combiner. The resulting signal then travels back to the 622-Mbit/sec receiver in the host digital terminal.
Active star network
The active star topology comprises the following parts:
Twenty 155-Mbit/sec medium-reach optical transmitter/receiver modules at the host digital terminal
A 48-fiber management area located at the terminal and equipped with associated jumpers, splicing, interconnect and storage hardware
A 48-fiber, armored, buried sub-feeder cable with 40 active fibers
A splice enclosure at the serving-area interface
One 155-Mbit/sec medium-reach optical transmitter/receiver module at each optical network unit.
The active star host digital terminal is equipped with twenty 155-Mbit/sec optical transceiver units. These units are connected to the serving-area interface with 40 fibers of the 48-fiber cable. At the interface, pairs of fibers are spliced through to 20 optical network units; each unit contains a 155-Mbit/sec transceiver. The 155-Mbit/sec signal received at the unit is further processed for distribution of individual signals to each of 16 homes.
The reverse direction signals sent from the homes are combined into a 155-Mbit/sec signal in the optical network unit. The 155-Mbit/sec signal is then transmitted back through the serving-area interface to the host digital terminal.
The key cost differences between the passive optical network and active star topologies cover the number of fibers, the optical splitters, and the bandwidth, range and quantity of the optical transceiver modules. Because the passive optical network design supports four optical network units per fiber span between the host digital terminal and the serving-area interface, then 622-Mbit/sec optics are required for each span. The active star host digital terminal supports one optical network unit per span, allowing the use of cost-effective 155-Mbit/sec optics.
Another difference between the two topologies results because the sub-feeder in the passive optical network needs one-fourth the number of fibers required by the active star topology. However, the network also requires the use of optical splitters, hardware and installation costs not incurred with the active star topology.
Taking into consideration the costs of electronics, hardware, cabinets, fiber-optic cables, installation, construction and incidental items, the cost advantages become evident. The calculations indicate that the active star distribution topology is less expensive than the passive optical network distribution topology for sub-feeder lengths of less than 23,000 feet. The shorter the length, the greater the advantage for the active star topology.
The crossover point occurs at a length significantly longer than the average maximum carrier-serving area range of 12,000 feet. Therefore, if the fiber-in-the-loop host digital terminal is deployed within a conventional carrier-serving area, the active star distribution topology appears to be significantly more cost-effective than the passive optical network distribution topology.
This cost comparison concerns a passive optical network architecture supporting 622 Mbits/sec of bandwidth in both directions. If, however, the upstream bandwidth is limited to 155 Mbits/sec for economic reasons, the relative cost crossover point would occur at 20,000 feet, only 3000 feet less than the first comparison. This crossover point is, therefore, not a significant improvement, considering the possibility of congestion due to limitations in upstream bandwidth. However, this crossover point occurs at a sub-feeder length much longer (167%) than the conventional carrier-serving area range.
Other considerations
Whereas the choice between the two topologies is made clearer by the cost and reliability comparisons, other differences must be investigated. For example, if the host digital terminal is placed in the central office to support a passive optical network in a non-ring arrangement, it would take 40 fibers of the feeder cable to serve four distribution areas in the carrier-serving area.
If the host digital terminal supports an active star topology in a non-ring arrangement, then 160 fibers are needed. But if a ring structure is used to support host digital terminals located within the carrier-serving areas, then a four-fiber feeder cable is sufficient for either distribution topology. When the potential fiber savings of a ring is coupled with the survivability improvement of a ring, the advantage of locating the host digital terminal in the carrier-serving area, not in the central office, is affirmed.
Another consideration is that the failure of one host digital terminal optical module in the passive optical network topology affects 64 customers. A similar failure in the active star model only affects 16 customers. In addition, placing the host digital terminal into the carrier-serving area is consistent with current digital-loop-carrier deployment practices and procedures. This placement is not the case with other deployment options.
Finally, upgrades in the active star topology are possible on an individual optical-network-unit basis by replacing the 155-Mbit/sec optics with 622-Mbit/sec units, resulting in a four-fold increase in bandwidth. Upgrades in the passive optical network cannot be performed individually; they must be done in groups of four. More important, a four-fold bandwidth upgrade from 622 Mbits/sec results in non-weather-hardened 2.4 Gbits/sec. u
Bill McDonald is manager of product planning for broadband fiber-in-the-loop products, and Paul Langner is responsible for broadband fiber-in-the-loop system engineering at Fujitsu Network Transmission Systems Inc. in Richardson, TX.