Loop-jumping

CLECs lead the way in reconditioning loops for DSL. But copper involves some tricky testing considerations

History suggests that it is difficult for telecom service providers to teach an old network new tricks. Luckily for providers of DSL services, technology developments have greatly expanded the potential of the copper plant for carrying high-speed data services.

That's not to say that testing and prepping copper loops for DSL doesn't require a painstaking amount of work, though.

As DSL services roll out faster than ever, carriers face new challenges inherent to the very nature of the copper medium daily. Namely, the challenge is to re-engineer equipment originally built to handle the modicum of bandwidth needed for dial-up voice into an Internet access platform that businesses want today. As a result, the test and measurement sector, once relegated to field technicians in hard hats or to the battered, fluorescent-lit workbench in the central office (CO), is now an essential part of the DSL rollout.

DSL explosion

Several factors have contributed to DSL's popularity. Consumer demand for high-speed Internet access combined with a need for incumbent local exchange carriers (ILECs) to move data traffic off circuit-switched networks have spurred more DSL deployments, especially by U S West and Bell Atlantic.

Although there has been considerable debate during the last few years about DSL's overall viability in a market leaning heavily toward fiber and digital access, the sheer amount of copper in the ground - especially in major metropolitan areas - has all but resolved the argument. DSL is economical, efficient and workable. It's here to stay.

"DSL is a norm for telephony," says Irwin Gerszberg, division manager in AT&T's local network technology organization. In September, AT&T announced a sweeping initiative to bring broadband access to local markets by using DSL and cable modem platforms. Copper wire is as good as a natural resource, Gerszberg says. "We've got 200 million copper pairs out there, each one capable of handling multimegabits of data."

But it really has been the competitive LECs (CLECs) that are behind the DSL explosion. As alternatives to ILECs, they have had no compunction about using DSL to target business customers with a far less expensive alternative to T-1 lines.

For all types of carriers, the DSL testing process generally is divided into loop pre-qualification, which measures the suitability of the physical copper plant for DSL, and turn-up verification, which determines how well the data service is working. Loop pre-qualification is not difficult, but it can be time-consuming and - if a carrier finds that the loops in question need extensive conditioning - could affect the rollout of service and the specific flavor of DSL offered.

Guilt by association

Copper has a bad rap is because it has been around a long time in an era when conventional wisdom says any technology that pre-dates the microchip is obsolete. The bandwidth capabilities of copper never were in doubt because when the public network was built, the only application was voice. Hence, no one ever thought the copper medium was going to be used for anything else. The whole network was built to optimize a service that used a mere 4 kHz of bandwidth.

DSL, on the other hand, is all about getting the full capacity potential out of the copper loop, raising that bandwidth to at least 400 kHz and, depending on line conditions, up to 1.1 MHz.

To boost the bandwidth, DSL signals transmit information at higher frequencies. Herein lies the rub, because the higher the frequency, the more susceptible it is to interference. In DSL's case, this can be from such ephemeral sources such as AM radio.

The other principal limitation is that signals over copper attenuate faster than they do on other media such as fiber or microwave radio. This is why asymmetrical DSL (ADSL) comes with the 18,000-ft caveat. Anything beyond that and the signal loss, a function of resistance (measured in ohms, or omega), results in too much lost information.

So the two major test and measurement concerns in pre-qualifying a loop for DSL are the amount of interference and the amount of loss.

In the testing process, this means accounting for several factors to make sure they fit within the required parameters (Table 1). "There's a lot more to DSL than just pulling a copper line," says Rex Humston, vice president and chief operating officer for Jato Communications.

"You've got to test if the copper is good enough to physically carry [DSL service]," says Peter Shaw, marketing manager for the service test division of Agilent Technologies. "There's not a lot of empirical data yet. Everyone is feeling their way in this new environment."

"From the global perspective, every telco we talk to has a different approach [to DSL line testing]," he continues. Some will contract out testing. Others will be extremely thorough to the point of putting a modem on each copper pair in a given cable and testing it. "There's a huge range of different approaches," he says.

DSL killers

But certain specific physical factors can rule out a copper loop from the start:

Digital loop carriers (DLCs). DSL requires a copper line from the main distribution frame in the CO to the point of entry at the customer premises. Even in the age of fiber, this is not as difficult as it might sound. Most leased T-1s are copper from end-to-end so the facilities are there. But DSL currently can't run through a DLC. Very high bit-rate DSL is designed to address that.

Distance. The practical limit for an ADSL loop is 18,000 feet - approximately 3.4 miles - from the CO. For high bit-rate DSL (HDSL) and symmetrical DSL the limit is even less - 12,000 feet. Resistance also can vary with the age and wear of the copper, so some copper loops are only able to handle DSL over shorter distances.

In the Northeast, where populations tend to be more dense and there are more COs, distance is less of a problem than in the Western states. "U S West tends to have long lines. Nynex [now Bell Atlantic] has short lines," Shaw says. On the other hand, plant in the East tends to be older than in the West.

For CLECs, CO location is critical in targeting customers. This reduces risk upfront, says Jato's Humston. Both the distance and DLC issues enter into the decision. "You want the right COs. Then you want to look at your customer's locations carefully. Fortunately, in cities like Denver, there's a CO generally a block away. And three blocks away, there may be another," he says.

Load coils. Signal loss due to attenuation was bad enough that in the early days of copper buildout, load coils often were added to prevent deterioration of signals carrying higher-pitched tones, mainly the voices of women and children. These load coilswere among the components that would come back to haunt carriers deploying DSL. Simply put, load coils and DSL don't mix. The load coils have got to go. Unlike the DLC cabinets that can be catalogued and spotted, or distance, which can be measured, often it takes test equipment such as a time domain reflectometer to find and isolate load coils because in many cases, records are unavailable or wrong.

Bridged taps and splices. Another physical aspect of loop architecture that affects DSL is the bridged tap. This is an extension, or bridge, added to a copper loop - in most cases to a new building or housing development. Unlike load coils, DSL can accommodate bridged taps, but only within certain cumulative limits. The problem here is the same with load coils: Carrier information dating back 50 years is prone to compound errors. The only way to find bridged taps is through line testing.

Beyond the physical obstacles lie the routine issues of electrical, electromagnetic and RF interference to which any transmission is subject. Aerial wires are more vulnerable than buried plant. However, one interference problem discovered recently has been so-called "power spectral density," which occurs with DSL loops that are grouped within the same cable bundle as leased T-1s. If the actual copper wires are too close to each other, it tends to create interference. The problem is more acute with ADSL, and the area of frequency overlap tends to be about 770 kHz, says Rick Puckett, product manager of telephony products at Tektronix.

For now, the spectral density problem must be dealt with by careful allocation of the copper facilities within each bundle, but techniques contained within the new HDSL2 specifications will deal with the problem electronically, says Taufique Ahmed, product marketing manager for the DSL products group at Level One Communications.

Then it gets tougher

When it comes to most of the physical problems that are part and parcel of DSL prequalification, off-the-shelf protocol emulators, spectrum analyzers and line testers from traditional vendors such as Agilent, Tektronix, TTC and Sunrise Telecom usually can get the job done.

But matters become more critical in the turn-up verification process. It's here that service providers, after qualifying their loops, determine if service is being delivered and if it's meeting quality goals. Routine testing here includes trace router testing, which maps out route paths and confirms connectivity to ISPs; LAN/WAN pinging, which measures network response time; and FTP rate tests, which measure delivery of fixed-size files from remote servers and calculates estimated packet arrival rates, packet error rates and circuit throughput (Telephony, Aug. 10, 1998, page 54).

Because this is where carriers compete and have the most control, test equipment for this phase tends to be more proprietary and more complex.

Loop prequalification, by its nature, requires a degree of fieldwork. But after that work is complete and the DSL is "lit," as the parlance goes, the goal is to keep truck rolls to a minimum.

Most CLECs maintain just one or two network operations centers (NOCs) from which they view their network operations in many cities. Most want equipment that can do remote testing and measurement and fix problems from afar. Jato, for example, uses Hewlett-Packard's OpenView management platform and hardware and software from companies such as Concord, Lucent Technologies and Micromuse.

In addition, the design and function of DSL enterprise networks, many of which carry only data, are driving changes in testing equipment and procedures. Most telephony test gear has been built for circuit-switched networks. Most CLEC networks, though they co-locate facilities in an incumbent CO, bypass the switch (see sidebar on page 44). The typical architecture is end-point to main distribution frame (MDF) to DSL access multiplexer (DSLAM) to IP router or carrier toll switch (Figure 1).

Managing DSLAMs

The introduction of DSLAMs into carrier networks has created new demands for test gear and given riseto some new companies.

Turnstone Systems, for example, is marketing a device that sits between the MDF and the DSLAMs and facilitates a number of testing and provisioning functions. Jato is one of many CLECs using the equipment.

One of the key functions of Turnstone's CX100 Copper CrossConnect is its ability to aid technicians in isolating specific copper lines (Figure 2). In the voice world, carriers have used mechanized loop testers and butt sets to find a specific line in the field. The technician knew when the right line was found because a dial tone was heard. However, because they connect directly to DSLAMs, DSLs don't transmit dial tones. So when a line problem is reported, locating and isolating the specific copper pair becomes a matter of trial and error. In this case, DSL providers can use the CX100 to send a signal down the line.

But the real utility in a device such as Turnstone's is its ability to automatically configure and provision DSL service from a bank of DSLAMs and document those changes in customer and carrier databases.

"In the case of a line card failure, we can reroute the subscriber to a different DSLAM card without sending a technician," says Eric Andrews, vice president of marketing at Turnstone. "Today that's a costly patch panel fix, and more often than not, the carrier doesn't have staff on site."

In addition, should a customer site require additional DSLs - or migration of service from an DSLAM based on ISDN DSL (IDSL) to one based on HDSL, the Turnstone equipment will make the change on command from an NOC anywhere in the country.

"Everything we do is managed centrally," says Jato's Humston. "There are a lot of tools involved. Turnstone is especially strong in the installation phase. Troubleshooting, too. Then we can log the historical characteristics of the loop."

The equipment's ability to simultaneously update several databases pertaining to outside plant is one of its most valuable attributes, he says. Indeed, it generally is agreed that information pertaining to copper loops and facilities outside the CO are among the most error-ridden in carrier logs. Most analysts agree that there is a 1-in-3 chance that data on any given carrier loop will contain inconsistencies or errors. The CX100 helps drop this rate by logging every change made to the facilities as it happens and sharing that information across databases.

The Turnstone equipment could be the first in a new line of DSL test gear specifically designed to be used with DSLAMs. Service providers say this equipment, for the most part, is sorely needed, especially when CLECs are trying to manage co-location and turn up service in hundreds of COs over the course of a few months.

Ultimately, the goal comes down to providing top-quality data service over plant built for narrowband voice. During the past few years, this notion has gone from questionable to doable. Although DSLAM equipment and new chipsets have contributed, service wouldn't be possible without the means to investigate and measure the effectiveness of individual loops to handle service. With much persistence, today's CLECs are showing that old networks can learn some high-speed tricks, even when those networks are made of copper.

Only three flavors reside at the very top of the DSL hierarchy, namely asymmetrical DSL, high bit-rate DSL and very high bit-rate DSL.

Originally envisioned for residential services such as video-on-demand, ADSL, which offers up to 9 Mb/s downstream and roughly 784 kb/s upstream, is proving a suitable method for consumer Internet access. Its main technological competitor is cable modem service, which poses its own set of testing issues (see story on page 54). Rate-adaptive DSL and the G.992.2 DSL specification, also known as G.lite or ADSL Lite, derive from ADSL.

HDSL, on the other hand, is proving attractive for enterprise users because it offers all the utility of a T-1 line at a fraction of the cost. Competitive local exchange carriers (CLECs) have seized on HDSL for this reason.

"ILECs are looking at DSL as a consumer play," says Shahim Bakhshandeh, vice president of engineering at NorthPoint Communications. NorthPoint and other CLECs are targeting small and medium-sized businesses that use T-1 and fractional T-1, where "$100 to $200 a month for DSL vs. $700 to $800 for T-1 gives DSL a huge advantage," he says.

HDSL2, ISDN DSL, symmetrical DSL and the upcoming HDSL2 are derived from HDSL. The last two often are set up when a copper line falls short of HDSL qualification.

The third flavor, VDSL, is seen as a likely evolution for service but, unlike ADSL and HDSL, is designed to operate over short copper loops. There is no widespread deployment at this time, and it has not begun to affect testing issues.

As they build out, competitive local exchange carriers want to test every piece of new equipment they add to their networks. Given the pace of change and the high demands and expectations of their customers for 24-hour service, all the major CLECs maintain labs with "captive loops" on which new equipment can be tested.

"It seems like we see a new DSL [access multiplexer] every week," quips Irwin Gerszberg, division manager in AT&T's local network technology organization. Exaggeration aside, Gerszberg estimates the AT&T DSL lab in Floral Park, N.J., examines a new DSL platform every four to five months.

In a typical lab environment, the equipment is put through a barrage of tests to determine how well it will hold up under real-world conditions. Although a great deal of standard equipment exists for the routine tests, to be assured of the network performance they want, most service providers say they end up taking off-the-shelf test equipment and reconfiguring it.

"There are a lot of things we need that don't exist today," Gerszberg says.

AT&T modifies a lot of test equipment for its own lab work. In addition, it developed its own ways to simulate traffic loads on an asymmetrical DSL link. "We try to mimic what users might do," Gerszberg adds.

NorthPoint Communications does the same thing, says Shahim Bakhshandeh, vice president of engineering at the CLEC. "What's not there is a tool that makes it easy to look at the [DSL network]."

As demand for DSL increases, this may change. New vendors, with gear designed for the broadband world, are emerging. In addition to Turnstone Systems, there are companies such as HarmonyCom, which offers equipment that manages and provisions broadband access services, including DSL, off various backbone platforms such as IP, ATM and Sonet.

When a user orders loops, we run different tests," says Shahim Bakhshandeh, vice president of engineering at NorthPoint Communications. "We check DC voltage, AC voltage, resistance, capacitance, load coils."

Because incumbent LECs (ILECs) sometimes operate with different sets of rules, "a [NorthPoint team] works with the ILEC on the interpretation of our standards and their standards," Bakhshandeh says. "There's no single solution end-to-end."

Jato Communications has a "joint handoff" process with the ILEC. After the line is ordered, it is flagged for DSL by the ILEC. It will be checked immediately to make sure there are no load coils or digital loop carriers on the circuit. The joint handoff is accomplished with technicians from the ILEC and the CLEC. Testing is done on the line at the moment of handoff. If the line doesn't qualify, everyone immediately knows. Problems can be identified and dealt with.

"The troubleshooting has been really successful," says Rex Humston, vice president and chief operating officer for Jato. The cooperative joint handoff process, he says, makes DSL pre-qualification much less of "a guessing game."