Here's a look at the important issues facing mass deployment of digital subscriber line services

Interest in high-speed access over digital subscriber lines is at an all-time high. Announcements of splitterless technologies designed to eliminate truck rolls, as well as a highly touted initiative by Compaq, Intel and Microsoft to develop a universal asymmetrical DSL splitterless standard, have only heightened the excitement.

Although the universal standard concept is powerful, we need to understand the breadth and depth of issues that have contributed to the delay in widescale DSL deployment.

Splitterless operation. DSL's ability to operate over the same copper wire that provides basic service is essential for such deployment. Initial designs required a splitter to separate the data service from the phone service at the customer premises and the network service provider location.

Emerging splitterless solutions will allow POTS and DSL data service to extend directly into the customer premises over existing intrabuilding phone wire. In many cases, the separation of POTS from the data and vice versa may be achieved with in-line phone filters, which provide filtering and separation directly at the phone jack where each phone is connected (Figure 1).

Although splitterless operation is critical, it is just one of several technical challenges that must be resolved to achieve mass-market deployment.

Loop reach and speed. Loop reach is one of the major determinants in market coverage. The longer the loop over which a given DSL operates, the greater the number of end users that each central office serves.

The use of higher frequencies for higher speeds results in decreased loop reach and service to fewer customers. This is due to a combination of the effects of high-frequency attenuation, bridged taps and crosstalk disturbance from other services.

For example, a 6 Mb/s ADSL with a 640 kb/s upstream channel uses frequencies approaching 1 MHz and typically has a loop reach of about 7000 feet of 26 gauge wire (Figure 2).

Unfortunately, nearly 60% of U.S. subscriber lines are greater than 7000 feet from the CO and therefore ineligible for this service. By comparison, a service that operates over 18 kilofeet extends the addressable market to 85% to 90% of subscriber lines.

Splitterless DSL technologies, which operate at frequencies below 100 kHz, have demonstrated speeds of more than 512 kb/s upstream and downstream at 18 kilofeet and beyond. However, estimates for the lower-speed splitterless ADSL.Lite, the proposed DSL standard that is projected to use frequencies up to 500 kHz, is expected to provide speeds of 96 kb/s to possibly 256 kb/s downstream, with 32 to 128 kb/s upstream.

This raises another issue-whether end users will be satisfied with high-speed download capability when they also need high-speed upstream capacity to upload to the corporate LAN to support videoconferencing at 384 kb/s in both directions.

So although market coverage is optimized, the greatly reduced speeds limit the services that can be offered.

Spectral compatibility. Electromagnetic waveforms transmitted down phone lines also radiate to other nearby copper wires. When they combine with signals running over adjacent lines, they can cause distortion and errors. This phenomenon is referred to as crosstalk and serves as an example of spectral incompatibility.

Although ADSL has been designed to manage crosstalk into other services, it still requires cable management to assure that ADSL and T-1 are not served in the same cable binder.

Similar issues exist with 2B1Q-based symmetrical DSL (SDSL) services, which interfere not only with T-1 and ADSL services but also with other SDSL services. Recent studies confirmed that crosstalk issues can be reduced dramatically and cable management virtually eliminated if the frequencies that DSL uses are limited to below 100 kHz and at lower transmit power levels.

Digital loop carriers. Virtually every telco has deployed digital loop carriers to support new lines and to reduce the effective length of long loops. A long-standing assumption has been that the industry simply will design ADSL cards to plug into those DLCs.

Although this has been demonstrated technically, it has proved to be highly impractical with existing ADSL technologies. The major issue is that the installed base of DLC remote terminals is extremely limited in space and power.

Full-rate ADSL equipment typically dissipates between 3 and 8 W per port and is fundamentally incompatible with the typical DLC power allocation of roughly 1 W per card slot. Work now underway to specify an ADSL.Lite standard should reduce the power slightly and may make it feasible to integrate into new third generation platforms. But it is unlikely to be sufficient to support practical integration into existing DLCs.

With a growing realization that 30% to 60% of the market for DSL services may be served through DLCs, this will be a key factor in limiting market coverage and affecting the overall business case. Fortunately, some vendors are implementing low-power technologies that address both the splitterless requirements and the DLC requirements.

Loop qualification and testing. Considering how loop length, crosstalk and DLC affect or limit DSL deployment, it's apparent that network service providers have their hands full in identifying which loops are qualified to serve specific customers.

In theory, telco databases should provide the information necessary to support initial screening for loop qualification. In reality, these records rarely exist, and if they do, they are often inaccurate-rendering the data useless. Some new DSL products are designed with programmable digital signal processors capable of providing full test equipment loop qualification and testing to alleviate this issue. In some instances, these products will provide loop qualification data even when only the CO modem is connected to the line.

These advancements, combined with efficient solutions that limit frequencies to below 100 kHz, will result in broader percentages of loops being qualified for the service.

Power dissipation of 4 to 8 W per port represents a challenge not only for the DLC but for COs as well.

Power dissipation and port density. Power dissipation is so significant that it can actually limit the number of ports that can be equipped in a cabinet and a CO. For example, an ADSL access multiplexer that has a nominal power dissipation of 4 W per port will dissipate more than 800 W of energy with just 200 ports equipped in a cabinet. The cabinet may be able to physically support more than twice that number of ports, but industry specifications governing the maximum heat dissipation become the limiting factor.

To put this into perspective, consider that a competitive local exchange carrier securing a 10 x 10 foot cage within a CO may be limited to a maximum heat dissipation of 800 W for the entire cage. This means that the CLEC could serve 200 full-rate ADSL customers while staying within specifications.

Fortunately, some splitterless DSL technologies provide dramatically lower-power dissipation with the potential to operate below 250 milli-watts average power. That would let the CLEC support more than 3000 subscribers from the same cage.

Home wiring and support for multiple PCs. With one exception, splitterless DSL technologies provide a point-to-point service that extends high-speed Internet access or remote LAN access service to a single phone jack with one or more PCs that can be attached locally. However, if PCs are located in different rooms throughout the customer premises, rewiring with Category 5 wire is required to support intrapremises LAN distribution of the network service. This is technically feasible but highly impractical in a mass-market model.

Some vendors have been working to develop solutions that convert the in-house wire to act like a LAN. But these solutions typically are incompatible with splitterless DSLs, which need to operate over the same in-house wiring. One recently introduced splitterless DSL product supports communication with and between multiple end points, which enables existing in-house wire to support LAN-like communication between PCs while supporting concurrent DSL communication over the local loop. This approach provides support for multiple PCs and eliminates the need to rewire the premises.

Multiple users, multiple solutions The network model. Dial-up modems use point-to-point protocol (PPP) to establish logical communication between the PC and the server at an Internet service provider or remote LAN. This has worked well for single user sessions, and standards groups are working to develop a similar network model with PPP over an asynchronous transfer mode virtual channel for DSL.

This effort may be well suited for single-user connectivity, but it is still evolving to support multiple users in a secure and efficient manner.

Some vendors also have developed secure multi- and single-user solutions with Layer 2 network models, which leverage virtual LAN architectures to provide scaling, security and multiuser support (Figure 3).

These solutions have the advantage of leveraging existing security and authentication that ISPs and corporate intelligent services managers have used for years.

While most business-centric CLECs have focused on virtual LAN type network models, many Bell regional holding companies with a more residential focus have a bias toward PPP over ATM. The result is that we are likely to see more than one network model address the varying needs of a wide range of end users.

The need for multiple network models may prove challenging for some network service providers and may provide a few wrinkles in mass-market deployment.

OAM&P. Service providers have the challenge of interfacing with legacy systems that integrate operations, administration, maintenance and provisioning and billing on an end-to-end service basis.

Although they can be managed on a moderate scale, the systems required to provide a positive customer experience when ordering and receiving the service are extremely challenging and will prove to be a key factor in mass deployment.

Cost. Anyone who has surfed the Internet knows there is value in driving access rates to the Internet or remote LAN from close to 56 kb/s to more than 500 kb/s. The questions are, how much is the market willing to pay, how much will it cost to overcome the complete range of deployment issues, and does the mass consumer service business case work?

Typical projections suggest end users will pay an average of $25 to $75 a month for both access and Internet service. The closer you get to $25, the greater the take-rate and vice versa.

Virtually all of these issues affect one-time and ongoing operational costs. They will help reduce costs and improve the business case's viability.

Work being done by the International Telecommunication Union and the Universal ADSL Working Group represents a good effort to address the need for a splitterless solution and is intended to drive mass-market deployment.

But the industry only will achieve this objective by resolving the broader technical and business issues. In the interim, many progressive network service providers are likely to achieve significant DSL deployment with more selective rollouts.