First-Mile Ethernet Camps Dig In
The group is meeting this week in New Orleans to work on a standard that will include definitions for first-mile Ethernet service over existing copper infrastructure. Currently, the working group is split into two camps. On one side are ADSL (asymmetric digital subscriber line) equipment and chip vendors advocating the use of a single physical layer (PHY) standard for both short-reach and long-reach applications. On the other side are their potential customers -- North American incumbent carriers and a few other equipment startups, which insist two PHY layers are needed.
"It’s really amazing how these vendors are thinking,” says Roy Bynum, an independent consultant, who used to work for WorldCom Inc. (OTC: WCOEQ). "It’s like they’re saying ‘We will have to educate the service providers on what we think they need.’ The level of hubris is unbelievable."
While the two groups have been able to agree on a few things, they continue to remain divided over one key issue: defining the long-reach Ethernet-PHY for copper applications. During the last meeting in July, they managed to agree that VDSL (very high bit-rate DSL), which supports higher data rates over shorter distances, should be used to fulfill the short-reach objective for the standard, which is 10 Mbit/s at 750 meters. But they cannot agree on which technology to use to fulfill the long-reach objectives -- 2 Mbit/s at 2,700 meters.
Vendors supporting ADSL for long-reach include Advanced Digital Information Corp. (Nasdaq: ADIC), Alcatel SA (NYSE: ALA; Paris: CGEP:PA), Aware, Broadcom Corp. (Nasdaq: BRCM), Centillium Communications Inc. (Nasdaq: CTLM), Ericsson AB (Nasdaq: ERICY), (Nasdaq: GSPN), IBM Corp. (NYSE: IBM), Ikanos Communications Inc., Intel Corp. (Nasdaq: INTC), Metanoia, NEC Electronics Inc., STMicroelectronics NV (NYSE: STM), Texas Instruments Inc. (NYSE: TXN), and Zarlink Semiconductor Inc. (NYSE/Toronto: ZL).
This camp holds that enhancements to ADSL, which is already used for most residential DSL deployments, can also be used in long-reach business applications. These vendors point to ADSL’s already installed base as proof that the technology works. The broad range of vendors making ADSL gear lowers the cost of equipment, making the solution inexpensive to deploy.
They also point to ADSL’s compatibility with VDSL technology. Both ADSL and VDSL can use discrete multitone as a line-modulation technique (see Ethernet Over Copper: Now You’re Talking). This would allow for product interoperability and easier installation. Metanoia's Daun Langston, one of the co-authors of a presentation advocating ADSL as the long-reach technology, says that using ADSL for long reach would greatly simplify the whole standard. He argues that this would allow for a single modulation technique to be used in the physical layer.
”802.3 Ethernet is something that is ubiquitous,” Langston says. “It’s cheap and always works. This is what people think about when they think Ethernet."
But incumbent carriers in North America like Bell Canada (NYSE/Toronto: BCE), BellSouth Corp. (NYSE: BLS), Qwest Communications International Inc. (NYSE: Q), SBC Communications Inc., Sprint Corp. (NYSE: FON), and Verizon Communications Inc. (NYSE: VZ), argue that ADSL is not optimized for business customers.
They want a service that strictly focuses on data transmission. Instead of ADSL, they propose using an enhanced DSL technology already defined by the International Telecommunication Union (ITU), called G.SHDSL. SHDSL’s biggest benefit is its long reach. It can transmit data over 20,000 feet or 6,000 meters. But it’s limited in its transmission speed to a range between 192 kbit/s and 2.3 Mbit/s.
Several equipment companies support the carriers, including Actelis Networks, Extreme Networks Inc. (Nasdaq: EXTR), Hatteras Networks, Infineon Technologies AG (NYSE/Frankfurt: IFX), Metalink Ltd. (Nasdaq: MTLK), and Symmetricon
ADSL technology can transmit data up to 3,600 meters. According to the vendors supporting this technology, this distance covers 95 percent of digital carriers loops (DCLs) and 85 percent of business and residential customers. But carriers say that they need asymmetric transmission, optimized for delivery of voice, video, and data in residential applications. And for business customers, the EFM/Cu PHY should be symmetric, and optimized for data services only.
While both groups seem to have viable technical arguments for advocating one approach or the other, underneath the techno-babble the real divisions seem to be political.
ADSL system vendors like Alcatel or chip vendors like IBM and Zarlink are protecting their turf. These vendors have already sold thousands of ADSL modems and chipsets into the marketplace. Introducing a new technology would mean that their products would have to be reengineered.
Dong Wei, of SBC, says the ADSL vendors are using this dispute to help them score points in the debate on modulation signaling standards between QAM and DMT. If they can get the standards body to adopt DMT modulation as part of the standard, they have won another battle over QAM, which is often used in VDSL applications.
Even so, it seems strange that vendors would deliberately build products that potential carrier customers say they don’t want. But Metanoia’s Langston says it’s only the U.S. carriers that seem to have a problem with their proposal. And he says that market is small in comparison to worldwide DSL deployments.
“People have lost their shirts trying to sell gear to the RBOCs,” says Langston. “Asia is where the money is. The U.S. market has been a disaster. Seventy-five percent of the population has access to DSL in Korea. In the U.S. less than 5 percent have access.”
But ADSL vendors might not be the only ones protecting their turf. Many RBOCs have a vested interest in the success of G.SHDSL. SBC, BellSouth, Qwest, and Verizon -- just to name a few -- have all been seriously testing G.SHDSL gear. The technology has been viewed as a replacement to traditional T1 service, the RBOCs' cash cow. Now, as many of these carriers are set to deploy G.SHDSL service, there is talk in this working group of changing the standards, an event that could render such G.SHDSL deployments obsolete.
Langston also claims that support within the RBOCs is not as united as it appears in the working group proposals. He says there is a lot of dissention among these carriers. SBC's Wei, who seems to be spearheading the RBOCs' efforts on this standard, denies that claim. While the IEEE is designed to hear the voices of individual engineers and not necessarily companies, Wei says that he has the full support of his company in the drafts he is proposing.
As the working group enters its third day of meetings in New Orleans, no decision has been made with regard to which technology will be selected as part of the standard for the long-reach application. Members attending the meeting don’t anticipate a decision being made until the next meeting in November in Hawaii. But some members of the group complain that they are running out of time and a decision needs to be made soon. It is pretty clear that the group will miss its original September 2003 deadline for a completed standard.
“It would be nice if we could have a single PHY for long reach and short reach,” says Wei. “But the reality is that given the time frame we are working with it’s unlikely that we will generate a technically sound solution that will cover both long reach and short reach. We just don’t have time to dig around into unproven technologies.”
— Marguerite Reardon, Senior Editor, Light Reading