Equipment vendors are discovering new ways to get more bandwidth out of copper access lines

Despite all the attention given to lightning-fast broadband speeds enabled by fiber to the home, most carriers are forced to concern themselves instead with how much bandwidth they can squeeze out of copper already in the ground.

Hoping to find profit in this pain, equipment vendors are working on a range of new technological advancements to push the envelope of what copper can do.

For example, Alcatel-Lucent introduced a new technique for bolstering copper data speeds this year. Dubbed “SmartDSL,” the technology is aimed at keeping noise levels on DSL lines low without compromising bandwidth speeds. When noise increases in a DSL network, the traditional solution is to rebalance power levels, increasing the signal-to-noise margin to compensate. But this results in less available bandwidth. Alcatel-Lucent's approach keeps the level of available bandwidth high by inserting two types of simulated noise: what it calls “artificial” noise and “virtual” noise. Artificial noise travels down the line itself while virtual noise is introduced directly at the receiver.

“By putting this noise in specific spots, you keep the signal-to-noise ratio high, increasing the speed and throughput of the line,” said Steve Kemp, senior director of product marketing for broadband access in North America for Alcatel-Lucent.

Kemp claims SmartDSL can potentially yield performance improvements as high as 30% above traditional DSL, but on average, improvements will probably fall between 10% and 20%, he said.

SmartDSL isn't available in North America yet, and Alcatel-Lucent isn't sure when it will be. To work, customer premises equipment (CPE) vendors must adopt the technology, Kemp said, and so far that hasn't happened. Part of the reason is the artificial and virtual noise combination — while the former is compatible with existing ADSL modems, the latter is an option within the VDSL2 standard and hasn't yet been applied broadly to industry silicon.

Other vendors are pursuing different ways of turning down the noise in (and thus increasing the power of) VDSL2. A group of companies that includes Israeli equipment vendor ECI Telecom is working on a technology called dynamic spectrum management (DSM) that promises to reduce the threat of cross-talk interference on VDSL2 lines, giving them greater reach and power. One aspect of DSM called vectoring coordinates the spectrums of multiple signals by processing them together so that cross-talk appears to lessen. In theory, vectoring potentially could double the speed of VDSL2 loops between 100 and 200 meters long, ECI Telecom said, perhaps extending the reach of 100 Mb/s service by 250 to 300 meters. But the commercial introduction of DSM vectoring must wait as standards are hammered out.

Another new equipment start-up, Phylogy, raised $1.4 million in Series A funding last year with a plan to extend DSL's reach. Phylogy claims its line conditioners can boost ADSL2+ lines to 12 Mb/s over 12,000 feet of 26-gauge copper.

Some of the more dramatic advances in copper are occurring at the chip level. A new vendor named Xtendwave is exiting stealth mode this year, promising to boost the speed and reach of DSL. Xtendwave's chip uses mathematical functions known as “wavelets” to modulate the time and frequency of signals simultaneously. Current DSL lines are based on fast Fourier transform, in which signals are defined by frequency but not by time. By using wavelets instead, and seeing signals in terms of both frequency and time, Xtendwave's chip has a clearer view of the line and is thus able to more efficiently pack bits into the pipe, the company said. The chip achieves this in part by overlapping channels instead of maintaining guard bands between them. Those guard bands are usually erected to maintain orthogonality (or multidimensional modulation), but wavelets are naturally orthogonal, Xtendwave said, so those guard bands can be replaced with more signal.

“By setting guard bands, you're determining the peaks and valleys of the energy cycle,” said Marc Landry, CEO of Xtendwave. “We're able to see [the signal] more cleanly and determine exactly where those energy peaks are.”

Armed with that heightened visibility and the greater efficiency it allows, Xtendwave claimed it can send more bits longer distances. Its chip can send traffic at T-1 speeds over more than 22,000 feet, the company said. At a quarter of that speed, it can reach more than 27,000 feet.

“Where the typical 40 Mb/s is stopping at about 3500 to 5000 feet, we're sending it 6000 to 8000 feet,” Landry said. Not only does that allow carriers to push triple-play services to areas previously beyond reach, it also boosts the bandwidth for subscribers already on the network.

But again, before Xtendwave's chip can work its magic, it has to be included in access and CPE alike. And the company has barely begun, hoping it will start shipping its product into commercial gear toward the end of 2008.

Earlier this year, another start-up chip vendor, Rim Semiconductor, announced that in tests it had sent traffic at 40 Mb/s over 5500 feet of 26-gauge copper wire. The company achieved this feat, it said, using a new technology it calls IP subscriber line (IPSL). The company is proposing an alternative to discrete multitone line coding, or DMT, the standard commonly employed in DSL networks including VDSL. Rather than reserve fixed allocations of upstream and downstream bandwidth, as DMT does, Rim's chip uses TDM — instead of the frequency-division duplexing used in VDSL2 — so that, when needed, downstream traffic can use bandwidth that otherwise would be reserved for upstream traffic, and vice versa. And it uses “rapid bidirectional switching” to transition in milliseconds from upstream and downstream transport. According to the firm's Web site, Rim's technology “defines” not just the encoding algorithms inside transport processors but also “the signal stream waveform.” Rim attacked rate and reach limitations in several ways, the company said, increasing payloads and decreasing noise and latency.

However, in order to work, IPSL would need to be embedded in both sides of the access network and be embraced by access and CPE vendors. To accomplish this, Rim has convened the IPSL Special Interest Group to create industry standards for the new technology.

Meanwhile, publicly held Rim is facing its own internal financial hurdles, reporting in January a need to raise more funding “immediately.” And it doesn't expect an ASIC-based version of its product to be commercially available until later this year.

Another promising means of boosting copper bandwidth is with line bonding — in particular, bonding the most advanced form of DSL, VDSL2. AT&T repeatedly has cited VDSL2 pair bonding in response to questions about whether its fiber-to-the-node (FTTN) architecture includes enough bandwidth to satisfy consumer demand, especially for multiple channels of high-definition television. But the goalposts for that technology have moved back some over time. In early 2007, AT&T talked about implementing bonded VDSL2 before that year ended, but as the year wore on, 2008 became the target for pair bonding. And in January an AT&T executive pegged “late 2008” for its introduction.

Alcatel-Lucent, AT&T's FTTN supplier, said its VDSL2 bonding gear is ready, but in order to be deployed it must be integrated into CPE.

Other vendors are taking different approaches to line bonding in an effort to boost broadband speeds for business customers. Along with vendors such as Actelis Networks and Hatteras Networks, add Mushroom Networks, an equipment start-up that launched this year. Mushroom's product, which is currently shipping, is a $3000 CPE device that bonds multiple lines of various types — cable, DSL, T-1, etc. — and combines their bandwidths into one virtual line. The box includes six ports on the network-facing side and four LAN ports facing the end user.

When a user wants to download a big file, for example, Mushroom's device sends separate requests for each network-facing broadband port — one for each section of the file, so that each section is downloaded simultaneously into separate ports without overlapping efforts, and the file is reassembled in the device. Thus, the user gets the benefit of the combined speeds of all the lines.

“It's almost like a deep packet inspection technology,” said Cahit Akin, CEO of Mushroom. “We look into [user download] requests and manage them intelligently and send them out through the DSL modem line. When it comes back to us, we do the reverse operation and present it as a single connection to the office network.”

Mushroom follows another new start-up with a focus on broadband bonding, British company Sharedband Networks. Unlike Mushroom's one-sided approach, Sharedband's system pairs CPE at one end with content delivery hubs in the network.

Sharedband's initial offering uses a bank of managed centralized servers, off-the-shelf routers from Netgear and its own firmware to bond up to four ADSL lines for greater bandwidth speeds. This month, the company also is adding another version using Linksys routers that allows bonding of non-ADSL lines, including those from different providers. The aggregation of lines adds a 3% overhead to the traffic, which the company says is no impediment to the quality of applications such as voice and video.

Another little-known company, Widearea Communication Technologies, promises to boost DSL signals with an ADSL loop extender and an “IPTV booster” that it claims extends the range of DSL by 30% to 50%.

Widearea's small, inexpensive gear is “refreshing in its simplicity,” said Kermit Ross, a consultant with Millennium Marketing — especially considering the mix of new technologies and new standards being developed to completely rethink how copper access networks are built.