The evolution of digital-loop carrier
SPECIAL REPORTS / Access Networks
With next-generation DLC systems' multi-interface and service capabilities, core and edge networks can evolve without expensive upgrades to the outside plant.
DARRELL MANSUR, Alcatel USA
Many forms of digital-loop carrier (DLC) are used in telecommunications OSP networks in North America. DLC systems support narrowband services at remote locations from telephone central offices, typically in locations where it costs less than extending or reinforcing copper-feeder facilities. Some systems, notably next-generation DLCs (NGDLCs) also support wideband and broadband services. Combinations of copper, fiber, and digital microwave systems are used for transport in between the central offices (COs) and the remote terminals (RTs).
DLC RTs operate using commercial power feeds. They also have battery backup and emergency generator outlets to support basic telephone services-lifeline plain-old telephone services (POTS)-in case of commercial power failures. The terminals consist of channel banks with slots for service and transport line cards. The service cards may each support an individual type of service or a combination of services, for instance, asymmetric DSL (ADSL) and POTS. Service features are determined by the shelf backplane capabilities and, with NGDLC, the associated system software. Each slot is hardwired to copper pairs that extend beyond the terminal, normally through protectors. The protectors prevent damage from power and lightning hits on the exposed plant beyond the RT. They are usually combined in 25-pair blocks, the factory stubs of which are most commonly spliced directly into the derived feeder cables originating at the RT. In some cases, feeder distribution interfaces are installed in the RT enclosure, in which case, the protector stubs are spliced to the feeder distribution interface in count stubs. The terminals may be housed in specially designed cabinets, equipment huts, controlled environmental vaults, or building terminal equipment rooms or closets.
The associated CO locations typically have CO terminals that provide interfaces to other equipment and, if the system has integrated optics, act as fiber-optic terminals. A CO terminal may support one or several RTs. The equipment is installed in relay racks and connected to the central-office power source.
In North America, DLC systems conform to American National Standards Institute standards and are rated against the Network Equipment Building Standard requirements. Those that interface with the legacy operational support systems used by the regional Bell operating companies also go through Telcordia's operation systems modifications for the integration-of-network-elements process to ensure compatibility with legacy operational support systems (OSSs) that perform inventory, provisioning, surveillance and maintenance functions.
With the advent of next-generation systems, older systems are often referred to as "traditional DLC." The earliest digital systems mimicked the D4 channel banks used for carrying narrowband services across interoffice facilities. Features such as RT capabilities were added to make them suitable for OSP. They were designed as "pair gain" systems, supplanting analog concentrators and carrier concentrator systems deployed earlier for that purpose.
In the United States, SLC-96 systems (and later, SLC Series 5) developed by Western Electric were the most widely deployed. They were based on T1/D4 technology where, in Mode I, 24 DS-0s (64 kbits/sec each) were transported over a 1.5-Mbit/sec T1 span. Each T1 span had a transport and a receive pair, so the nominal pair-gain ratio was 12 to 1. However, additional pairs were used on a system basis for communications, fault isolation, and span protection, reducing the actual gain to eight to one. Mode II doubled the nominal gain with 48 DS-0s served by a T1 span.
The first systems, introduced in the mid-1970s, were copper-fed, with five T1 spans used for each 96-channel system (including the protection T1). When fiber was introduced into the loop in the early 1980s, three-to-one multiplexers in external fiber-optic terminals were installed to generate the electrical T1 signals; the systems multiplexed DS-3 (44.736-Mbit/sec) signals into 28 DS-1s or T1s. (In this case, loop refers to the portion of OSP located between the CO and telephone customers, from the CO main distributing frame to the network interface device at the customer premises.) A notable exception was an integrated fiber-optic solution for SLC-96 that used a proprietary "fat fiber" (62.5-micron) interface. That was especially helpful for cabinet installations, whose space limitations often precluded the use of separate fiber-optic terminals.
In 1983, the demand for the SLC systems exceeded supply due to a worldwide shortage of silicon chips, opening the door for wider deployment of other systems with different schemes. The most popular of these in the United States was Nortel Networks' DMS1 Urban system. It supported up to 544 POTS lines with two to eight T1 spans, using dynamic time-slot concentration. This concentration scheme increased T1 span utilization and reduced the need for fiber.1
Both companies, Western Electric and Nortel, offered proprietary switch integration options for their systems, reducing the need for narrowband line cards in the CO terminals. These configurations were referred to as integrated DLC. System configurations requiring CO terminal line cards became known as universal DLC. Integrated DLC was initially hampered by the proprietary interfaces and the need to emulate SLC-96 in the legacy OSS. Its use dramatically increased with the adoption of switch interface references (TR-008 and -003)2 and the associated development of NGDLC systems.
Another disruptive influence near the end of the traditional DLC's reign that helped bring in NGDLC was the introduction of ISDN. Although it never gained the popularity in North America that it achieved in Europe, the Integrated Services Digital Network (ISDN) changed DLC system requirements and even the way DLCs were deployed. The introduction of the carrier serving area (CSA) concept in 1982 also had an affect. It restricted serving areas beyond RTs and prohibited the installation of load coils in the derived feeder facilities.3 These guidelines supported ISDN and other narrowband digital services.
In the late 1980s, NGDLC systems were developed to optimize the use of the integrated switch interfaces and SONET transport, incorporating software control features. With improvements in component density, the systems could also scale to larger sizes to better fit CSA applications.4 The early lead was taken by the DISC*S system now supplied by Marconi. The Litespan system, introduced with integrated optics, later overtook that lead and is now supplied by Alcatel. In recent years, AFC's UMC-1000 platform has become popular for rural applications.5
Using TR-303/GR-303 reference de-signs, CO terminals are optional for these systems. However, CO terminals are necessary when integrated optics are used and are advantageous for combinations of integrated DLC and universal DLC. The latter continues to support non-switched special services and unbundled loops. GR-303 concentration, using time-slot interchange functionality, further reduces span requirements. With widespread fiber deployment, that benefit is now focused on improving switch-interface utilization.
As with traditional DLCs, the primary NGDLC application is feeder relief. The combination of integration and scale reduced average line costs, however, increasing NGDLC deployment. NGDLC overtook traditional DLC in the mid-1990s and moved the average economic crossover with copper feeder closer to the CO. Figure 1 depicts this application.
Another NGDLC application that has gained some popularity is fiber-to-the-curb (FTTC), following Bellcore's TR-909 reference guidelines. In this case, either a CO terminal or RT becomes a host digital terminal for optical line terminals and the associated optical-network units (ONUs). The latter are installed at the curb or premises and house the service line cards.
Although FTTC systems are useful for deploying alternative video networks and higher-speed data, they proved expensive for lifeline POTS. The ONUs require network powering and the power distribution systems tend to cost as much as traditional copper distribution cable systems.
Partly because of the FTTC costs, additional effort was put into getting more out of copper. That resulted in the development of DSL capabilities, initially deployed with CO-based DSL access multiplexers (DSLAMs), which have gradually been integrated into NGDLC.
The most popular DSL deployment in NGDLC has been ADSL. It is a predominantly residential service, and the majority of NGDLC lines serve residential customers. Various network schemes and ADSL technologies have been deployed, but most development has been focused on ANSI-standard discrete multitone ADSL, using ATM networks. Some of the NGDLC systems already had ATM capability. Others have been upgraded to support it. The primary transport options have been to combine the TDM and ATM traffic on a single transport facility or to use separate facilities. The first option makes more efficient use of the transport facilities. The second option protects the TDM capacity from the ravages of the variable data demand.
Significant wiring and cost advantages can be gained with integrated ADSL at remote NGDLC locations. That's especially true where there is potential POTS growth, since POTS and ADSL can share the same derived pairs and line cards (line sharing). Some traditional DLC systems also can be upgraded for ADSL, using special line cards and statistical multiplexers. In both cases, combination ADSL and POTS line cards can be used. They have integrated splitter functions, precluding the need for external splitters.
The demand for residential ADSL is strong, but by some estimates more than half of the residential locations are beyond the reach of CO-based ADSL. You would think the deployment of integrated ADSL in NGDLC would mirror that of CO DSLAMs to reach those customers, but that has not been the case. Only a small portion of ADSL deployment to date has been through NGDLC.
Incumbent local-exchange carriers (ILECs) cite regulatory uncertainty as a reason for not adding or upgrading ADSL-capable systems. While CO DSLAMs are excluded from unbundling, NGDLC systems are not. ILECs are concerned that they would have to share their integrated "advanced services" facilities with competitive LECs (CLECs) at prices below their actual costs. To put the situation in perspective, compare ADSL deployment to cable modems. The latter also provide high-speed access, but cable operators are not faced with the same unbundling issues as the telcos. ILECs serve over 95% of U.S. households, whereas cable operators serve about 65%. Nevertheless, cable operators have more than twice as many cable modems in service as telcos do ADSL.
Both the unbundling and pricing rules are under review. Current Federal Communications Commission (FCC) rules for sub-loop unbundling actually exclude advanced services equipment, except where remote collocation cannot be accommodated. However, other rules require line sharing, including fiber-fed facilities, and the FCC's Collocation Further Notice of Proposed Rulemaking (FNPRM) entertains options such as line-card plug and play in NGDLC systems. At least one state (Illinois) has stepped beyond the FCC and ordered the latter, even though it is unfeasible to install cards that are not made or licensed by the NGDLC vendor. Other states are considering similar rules. The FCC may withdraw or severely restrict line-card interoperability and a Tele communication Act amendment could exempt the equipment.6
In any case, it is clear the issue is delaying NGDLC upgrades for ADSL. The primary exception has been SBC and its Project Pronto. SBC is leading the way with ADSL deployment in NGDLC systems, helping it to increase its overall ADSL service lead while it fights the regulatory battles. However, even SBC backed off where it felt the regulatory burden was too onerous, halting Project Pronto in Illinois because of that state's unbundling rules.
The pricing issue deals with the wholesale pricing methodology known as total-element long-run incremental cost (TELRIC), which is supported by the FCC and CLECs. The U.S. Supreme Court will hear arguments for and against its use in October. There are no promised dates for an FNPRM ruling, a TELRIC decision, or additional legislation, so these issues are likely to drag on well into next year.
Other regulatory issues of note involve the provision of space for CLEC collocation equipment at NGDLC RT locations and access to the derived feeder pairs leaving the RT sites. Both of these issues can be resolved with intercompany planning.
One of the technical concerns deals with possible interference between CO and NGDLC-based ADSL services that share the same binder groups in distribution cables. Since most NGDLC systems are deployed beyond the normal reach of CO-based lines,7 the chance of such interference is limited, but spectrum management rules need to be developed to handle the incidences if they do occur.
Other technical questions that need to be explored deal with back-to-back NGDLC systems deployed, respectively, by ILECs in their OSP networks and CLECs in CO collocation spaces. All services that might be unbundled need to be reviewed and tested in this configuration because more than 50% of the ILEC loops will be served by DLC and NGDLC systems within a few years.
The regulatory and technical issues will be worked out in favor of mass deployment of ADSL in NGDLC systems. Future enhancements will include additional ATM-based services and an evolution to IP-based systems with Gigabit Ethernet transport.
Additional ATM service capability will most certainly include the recently approved G.SHDSL standards for business data services. Eventually, the even higher-speed G.VDSL standards may be deployed in NGDLC for multitenant-unit applications. These applications also will be data services for the most part. The business case for video on DSL will continue to be tough with cable and satellite competition.
As the demand for these services materializes, it will stimulate the need for increased bandwidth in backplanes and transport links. This, in turn, will support the development and deployment of IP-based, integrated access, multiservice DLCs. The IP functionality will evolve with the migration from circuit to soft switching in the CO as well as the IP evolution in the rest of the network. The new DLCs will perform "universal line frame" functions for the IP switches and networks.
During this transition, the new DLC systems will have to support traditional TDM, ATM, and emerging IP services. With their multi-interface and service capabilities, these systems are the best-positioned network elements to support this convergence during the evolution from TDM to packet. Core and edge networks can evolve without expensive upgrades to the OSP network.
As the evolution to IP progresses, there will be changing requirements for gateway and transport functions. Near-term requirements include inverse gateways that allow voice-over-DSL services to be delivered to TDM circuit switches. Standalone voice gateways are already being deployed. Their functions could also be integrated in NGDLC systems. The feasibility of that depends on the pace of IP switch migration away from circuit switching. A fast transition would obviate the advantages of such an option. Future IP gateway functions are more likely to be integrated in distributed DLC configurations. Initially deployed in CO nodes, these functions will eventually migrate to RTs and even home gateways.
Currently, both the TDM and ATM transport facilities used for NGDLC systems are supported by internal and external SONET elements. Eventually, the SONET transport functions will migrate to Gigabit Ethernet connections that interface directly with Gigabit Ethernet backbone and switch networks (see Figure 2).
Still deployed in CSA configurations, the RTs will also serve as host digital terminals for passive optical networks (PONs), initially ATM PON and eventually Ethernet PON. Most RTs will continue to serve embedded copper facilities as well as the new fiber distribution plant.
OSP facilities will evolve with new capabilities and features, supporting a variety of higher-bandwidth services and gateway functions that are seamlessly integrated into end-to-end networks. Massive displacements or overlays with new technologies are highly unlikely due to the magnitude of such investments. Today, OSP accounts for approximately half of the telephone plant investment. Yet, with IP, we have a solid new end state we can use for planning new systems and services, providing a common direction that can benefit both equipment suppliers and service providers.
Darrell Mansur is a senior manager at Alcatel USA (Petaluma, CA).
- Another system using a similar concentration feature was Ericcson's Timespan 128. Designed for lower-density rural areas, the system also allowed RTs to be chained, sharing the same T1 spans.
- These are "shortcut" versions of the longer Bellcore/Telcordia TR-TSY titles.
- In the 1964 AT&T OSP transmission design practices, loading was required for loops extending beyond 18 kft, including bridged tap. That was to offset attenuation and was based on an approximate, -8-dB non-loaded loop limit, at 1,004 Hz, using Resistance Design rules for the cable gauges. With DLC, an approximate -5-dB limit (and 900 Ohms) was used for non-loaded loops beyond the RT, due to system loss.
- CSAs were defined by the Bell system staff in Recommendation Letter 82-02-207 as non-loaded areas extending 9 to 12 kft from the RT, including bridged tap. Bridged tap was limited to 2,500 ft with no single tap exceeding 2,000 ft. Combined with the LROPP recommendations issued by Bellcore soon after the 1984 divestiture, CSAs were usually designed to serve two to four distribution areas for a total line count ap proaching 2,000 POTS and equivalents. The serving-area interface at each distribution area could be fed with a combination of loaded copper-feeder pairs from the CO (or non-loaded pairs if the design point was within 18 kft, including bridged tap) and non-loaded derived feeder pairs from the RT.
- Other NGDLC systems include Nor tel's AccessNode, Lucent's SLC-2000, Fujitsu's FACTR, Next Level Communi-cations' BDT/USAM combination, and NEC's ISC-303.
- One measure would be the Tauzin-Dingle bill, HR 1542, which recently passed its subcommittee review. The bill would exclude high-speed access equipment from unbundling, but it is facing stiff opposition.
- Downstream ADSL data at 1.5 Mbits/sec can reach 16 to 18 kft, depending on cable gauge, if there are no interferences (and the pairs are not loaded). Since there normally are interferences, including other ADSL lines, 12 kft is typically used as the maximum range without loop qualification and conditioning. NGDLC systems are normally deployed at this point and beyond for economic relief.