Whether a carrier or multiservice operator, broadband to every home and business is dependent on how deep fiber extends into the access network.
Dan Parsons
Advanced Fibre Communications
The increase in backbone network capacity through DWDM has produced a bandwidth rippling effect starting in the backbone, moving to the metropolitan area, and now to the local access network. The "last mile" for residential broadband services has been the "Holy Grail" for carriers, access providers, and vendors alike. This area of the network is the grass roots for revenue because it's where services are deployed. However, it is also cost-sensitive, environmentally demanding, and requires the most maintenance.
Residential broadband services are now available to many households, but not all. Traditional telecommunications carriers and multiservice operators (MSOs) are currently limited in broadband deployment because of the installed cable plant or the availability of the required technology. For the last few years, the MSOs have offered broadband by means of the high-capacity cable-TV coaxial-cable plant and the cable modem. Telecommunications carriers, on the other hand, are taking advantage of DSL technology to overcome the limitation of the copper pair to the home, which was originally intended strictly for voice communications.
In the long-haul portion of the network, capacity is king. Vendors focus on density and channel count per fiber. In the access network, link capacity is not nearly as important as the following attributes:
- Transparency, or the ability to support both time-division multiplexing (TDM) and data services.
- Scalability to quickly support demand and network growth.
- Dynamic provisioning to provide services for high-bandwidth applications.
A number of technologies and architectures are available for delivering Internet-driven broadband services over existing copper infrastructure. Cable modems over a hybrid fiber/coaxial (HFC) network provided the first residential broadband services. Now, asymmetric DSL (ADSL) over the existing copper telephone wire is beginning to take off. By 2002, it is estimated that 12.9 million DSL lines will be put into service, up from 770,000 in 1999, according to Piper Jaffray in US Bancorp's February 2000 report. It is expected that 60% of these lines will serve the residential market.
Voice services still provide the largest portion of a carrier's revenue, even though data traffic is beginning to exceed voice. For traditional cable-TV operators that use an HFC network, providing voice services has proved daunting. Conversely, traditional telecommunications carriers have been challenged to offer broadcast-TV services on their networks. The challenge is to provide an access architecture, upon which all services can be deployed, managed, and scaled cost-effectively.
As higher-speed services move deeper in the network, so does fiber-optic technology. HFC technology brought broadband to the home primarily by extending fiber deeper into the local access network and closer to the home. But the current HFC architecture is limited.
Service providers cannot meet the bandwidth demand as the "take rate" for broadband services increases. That's because of the shared cable architecture and the multi-access technology of the network. Simply put, more subscribers are now competing for the same bandwidth. To mitigate this, the network architecture must drastically reduce the number of homes passed in order to decrease the number of subscribers trying to access the same bandwidth.
ADSL services are restricted to subscribers living within 18,000 ft of the central office (CO). The area beyond the 18,000-ft boundary can represent one-third to one-half of a carrier's potential broadband subscribers. This area is also experiencing the highest demographic growth, according to some reports. The population of Philadelphia grew 2.8% from 1970 to 1990, yet consumed 33% more land, according to The Sprawl Watch Clearinghouse. Similarly, the San Francisco Chronicle reported last December that, according to SBC Communications, almost half of the San Francisco Bay Area cannot get DSL service because they reside too far away from the CO. San Francisco is one of the cities with the highest concentration of technology in the world, yet its residents are faced with the limitations of the local access network.
The access network needs an evolutionary rather than a revolutionary approach for carriers to address demand for broadband services and at the same time satisfy traditional plain old telephone service (POTS). More important than technology is the architecture of the access network and network planning. Architecture transcends technology. As technology is developed, good network architecture and planning provides a means by which technology can be deployed. Predicting growth in the area beyond the CO DSL radius can be difficult. Network architecture will help alleviate the cost of predicting incorrectly and provide a means by which higher-speed services can be deployed in the future.
The success of residential broadband in the HFC network can be greatly attributed to taking fiber-optic technology deeper in the network and closer to the home. Digital loop carriers (DLCs) have provided a means for telecommunications carriers to deploy narrowband services in an evolutionary, low-risk manner. It is a cost-effective way for traditional carriers to provide services beyond the reach of the CO. The DLC eliminates the cost of access network expansion. Moreover, network planning can be done without the financial risk of building COs for new subscribers.
To address higher-speed services and further reduce costs, integrated DLCs (IDLCs) provided another evolutionary means by which carriers could offer services using the same architectural approach as DLCs. The IDLC provided a means by which optical fiber could extend into the access space, integrate more functions of the CO's switch, and deliver higher-speed services such as Integrated Services Digital Network (ISDN).
However, traditional IDLCs do not directly address DSL. Currently, most DSL access multiplexers (DSLAMs) are CO-based and do not extend broadband services beyond the reach of 18,000 ft from the CO. Using the IDLC approach, a fiber-connected integrated multiservice access platform (IMAP) with DSL services now extends ADSL services to broadband customers that would otherwise be unreachable by CO DSLAMs. These remote-hardened IMAPs must address the challenges of the traditional DLCs and IDLCs and provide transparency of service, scalability, and dynamic provisioning, along with backward-compatibility, cost sensitivity, and network management, and still meet the harsh environmental challenges for roadside access equipment. In other words, an IMAP must combine the enhanced service capabilities of a multiservice DSLAM with the battle-hardened capabilities of a broadband DLC system and have fiber connectivity to the CO.
Currently, the technology of choice for IMAP is ATM, which provides a way for IP and non-IP services to coexist on a single platform. Also, ATM can provide multiplexing gain through statistical multiplexing, which can in turn be carried in a SONET payload-the transport used by the majority of carriers.
The most significant investment for a carrier is in the outside cable plant. Broadband demand is pulling fiber closer to the home. Consider a fiber-fed IDLC servicing a new subdivision of 500 homes with POTS. The fiber link is likely OC-3 (155 Mbits/sec) with a DS-3 (44.736 Mbits/sec) of bandwidth provisioned for POTS. Providing broadband services could be as simple as adding ADSL cards and an OC-3c ATM link to evolve the IDLC into a remote-hardened IMAP with TDM POTS services. Considering a 10% take rate in the subdivision for 1.5-Mbit/sec ADSL Internet services, the OC-3c link will suffice without any ATM statistical multiplexing. In fact, all 500 homes in the subdivision can be serviced with 1.5-Mbit/sec ADSL services considering a 20:1 concentration.
Is this same subdivision now ready for digital-video services over ADSL? The traffic dynamics for video and Internet are quite different. While bursty Internet traffic can be statistically multiplexed, video traffic is more deterministic and cannot have the same concentration ratio as Internet traffic. Also, IP Internet traffic can tolerate more dropped packets and delays than video due to the nature of the service. Dropped Motion Picture Experts Group 2 (MPEG2) video frames can result in an unusable service. Furthermore, MPEG2 video requires at least 3 Mbits/sec to provide a VHS type of quality for movies. With these parameters for video service and an Internet ADSL take rate of 10% with a concentration of 20:1, the same remote IMAP can provide up to 25 homes of digital-video service without upgrading the capacity on the fiber link. Higher broadband take rates can be accommodated if the remote IMAP can be upgraded to an OC-12c (622- Mbit/sec) ATM link. Depending on the copper plant and the speed of the current ADSL service, video services may also reduce the ADSL service area from an IMAP from 18,000 to 12,000 ft.
What if the subdivision expands or other subdivisions now want broadband services? A carrier will certainly want to minimize the outside-plant costs and utilize the fiber link from the CO to the initial IMAP rather than pull more fiber from the CO to other IMAP nodes. Figure 1 shows how that can be done. The initial IMAP provides the ADSL and POTS services to subscribers and becomes a concentrator for other remote IMAPs. As the local access network grows to accommodate new remote service areas or higher-speed services, the CO link capacity will increase. With the advent of very-high-speed DSL (VDSL) at 50 Mbits/ sec, this CO link could approach the OC-192 (10-Gbit/sec) level of capacity. Depending on the installed fiber characteristics and its length, an OC-192 link may not be possible.
A WDM approach for the network expansion would be more practical and cost-effective. Currently, DWDM technology using wavelength spacing of 50 to 200 GHz will not meet the harsh operating requirements of the access network of -40°C to +65°C. The difficulty comes in maintaining laser wavelength stability over the operating temperature. A coarser optical-channel-spacing WDM approach would allow laser wavelength drift and still provide the optical-channel density needed in the access network. To achieve the OC-192 capacity, only four lambdas are needed for transport of four OC-48 (2.5-Gbit/sec) links.
Furthermore, VDSL services take fiber even closer to the home-within 2,000 ft. As shown in Figure 2a, fiber can be deployed from the initial IMAP node providing ADSL to an optical-network unit (ONU) delivering the new VDSL service. A more fiber-efficient and cost-effective way of implementing the same network is shown in Figure 2b.
This fiber-to-the-curb (FTTC) scenario is part of the evolution to fiber-to-the-home (FTTH), which many carriers would like to have today. Now that fiber extends deep into the network, realizing FTTH from FTTC does not require a large leap of faith. By this time, cost-effective FTTH subscriber technology will be commercially available to support even higher speeds of communication when compared to VDSL. Given that the fiber-plant architecture is built upon the DLC evolutionary model, the transition to passive-optical- network (PON) technology is possible without a heavy financial burden on cable-plant deployment. Also, given that the network has evolved to FTTC, a carrier can migrate the technology in the initial IMAP node and the ONUs to incorporate the optical splitters for FTTH and still maintain services to FTTC portions of the access network. The IMAP node and the ONUs can eventually transition from a hybrid to a complete PON providing FTTH.
Having discussed how the architecture of the access can be laid out to address physical outside-plant growth, the network would become unruly and costly to maintain if intelligence through network management is not considered. Access network elements need to have the same management philosophy as metro and backbone equipment. Fiber is extending deeper into the access network and will eventually reach our homes. The evolutionary DLC-to-IMAP approach allows carriers to grow the network at relatively low risk, satisfy the demand for residential broadband, maintain traditional services, and maximize revenue potential by reaching every home in the access network.
Dan Parsons is a senior manager of product management at Advanced Fibre Communications (AFC-Petaluma, CA). He can be reached at [email protected].