[ Monday's News ] ..... Or .....
The secret compartment of my ring I fill with an Underdog Super Vitamin Pill.
This one starts out good. But it sounds like the Telco's will have to replace a lot of copper though. Probably an every day job ... somewhere.
It's got a spin I can't put my finger on.
Ray
Jun 20, 1997
------------------------------------------------------------------------
June 23, 1997, Issue: 959
Section: Communications Design -- Part 3: Crisis In The Local Loop
------------------------------------------------------------------------
Central office sees a rocky renaissance
By Loring Wirbel
Listen to semiconductor vendors or WAN-access OEMs, and you might be forgiven for thinking a host of new wideband and broadband access technologies are coming soon to a telco central office near you. They will enable a consumer market of analog and digital modems that could hit retail shelves by Christmas. Here are just a few of the "hot action" items:
- Pulse-code-modulated (PCM) modems, a new breed of analog modem capable of downstream data rates beyond 50 kbits per second, are moving toward a draft standard in the Telecommunications Industries Association (TIA). TV ads already have created an end-user furor.
- In the digital domain, T1 leased lines with bandwidth of 1.5 Mbits/s are getting cheaper, and channel processors to support multiple services over T1 are getting smarter.
- Waiting in the wings is the family of technologies known as digital subscriber lines, which use twisted-pair wiring in the telco's local loop to provide symmetric or asymmetric data services as high as 6 Mbits/s downstream. Now that scores of chip-set, board and system makers have pledged to work on both DSL client modems and access multiplexers, it appears to be only a matter of months before DSL hardware and services are available for every residential and business application.
Equipment that switches and concentrates different traffic types within a central office has gone through a renaissance of redesign in recent months. Switches and multiplexers are now expected to handle analog modem pools, ISDN interfaces, T1 and T3 lines, DSL lines of various types and native asynchronous-transfer-mode (ATM) transport services. Scores of new central-office access systems debuted at the recent Supercomm show in New Orleans, designed to fit markets ranging from nationwide interexchange carriers to small Internet service providers (ISPs)
But wait a minute. Remember the Integrated Services Digital Network? The dual 64-kbit/s B channels of basic-rate ISDN (BRI) were supposed to represent an ideal customer interface to new services, as well as a way to get regional carriers accustomed to digital services. Instead, BRI is a virtual non-starter in the United States, except in regions where incumbent local-exchange carriers (ILECs, a.k.a. Baby Bells) have some urgent reason to deploy it.
Though ISDN experienced early technical glitches-problems with Signaling System 7 software and confusion about service-provisioning IDs-the real disincentive was the dirty little secret of the copper local loop in the ILECs' outside plant.
Many rural local loops were far longer than the 18,000 feet from the central office most ILECs claimed-indeed, some copper pairs extended as far as 40,000 feet from a central office or digital loop carrier. And lines were far "dirtier" than anyone let on in public, with undocumented bridge taps and other inconsistencies. When customers requested ISDN service on a mature loop, the ILEC almost always had to send out a craftsman to analyze the quality of the individual line.
Yes, Toto, there is a crisis in the local loop.
"You have to realize the craftspeople talk about weeks to analyze one line for ISDN," said Mike Wodopian, vice president of communication markets at Advanced Micro Devices Inc. (Austin, Texas), "and they say it will be tougher for the DSLs-particularly asymmetric DSL. We've stayed active in subscriber-line circuits because the uptake is predictable. But having been through the ISDN wringer on technology push vs. applications pull, we remain very skeptical of the time frames talked about for DSL service deployment."
David Helfrich, vice president of marketing and sales at DSL startup Copper Mountain Networks Inc. (Mountain View, Calif.), would have every reason to minimize problems. Nevertheless, he reports that Pacific Bell craftspeople tell him that "what was true for ISDN becomes far more serious for DSL." Copper Mountain is lucky: "At least we are working with a mid-speed version of DSL based on 2B1Q coding, giving us the expertise base from ISDN," said Helfrich. "But this is a problem everyone in the DSL community will be facing."
As if line quality wasn't enough to hobble DSL, regulatory uncertainties following passage of the 1996 Telecommunications Reform Act make it uncertain precisely which type of service providers can offer DSL services. Many startups boast of selling DSL access multiplexers into concentration points as small as an ISP point of presence (POP). Midspeed DSLs that are concentrated on a leased T1 line could certainly be offered by an ISP or competitive LEC (CLEC). But high-speed services requiring true central-office termination, such as ADSL, are a different story.
ILECs are supposed to "unbundle" the local-loop physical plant, allowing ISPs and CLECs to install central-office equipment directly on the local loop. In some instances, this might put ISP or CLEC equipment physically on-site at an ILEC's central office. Analysts expect ILECs to drag their heels in implementing the unbundling procedures. Others fret that ILECs may open up the copper plant, but threaten to sue an alternative carrier if a customer's DSL service has a negative impact on anyone else's service.
****
That threat is real-particularly with higher-speed services. But Krista Jacobsen, senior engineer with Amati Communications Inc. (Sunnyvale, Calif.), insists that "a T1 line in a binder group would be far more likely to wipe out an ADSL service than vice versa." Twisted pairs are carried in clusters of 20 to 25 "binder groups," and the chances of spectral incompatibilities across a group rises with every additional DSL service in one cluster.
****
ISP split
Bobbi Murphy, senior broadband analyst at Dataquest Inc. (San Jose, Calif.), expects a two-way split among ISPs: Smaller ones won't want to build out their own cable plants or place equipment in a telco central office; they will focus on minimal dial-up services. A smaller percentage will build their own facilities and become de facto CLECs. That scenario makes a handful of ISPs potential purchasers of DSL equipment, to be sure, but it's a far smaller number than some OEMs are betting on.
"Keep in mind that the CLECs and ISPs also are frozen out of the digital loop carrier [DLC]," said M. Niel Ransom, vice president and general manager of access and network management at Alcatel Network Systems (Raleigh, N.C.). "Many loops never terminate in a central office, but in a DLC box in a remote site. It is not clear that an alternative carrier could ever get access to the DLC."
Ransom suggested that "it is also interesting to look at the case where an ISP just wants a slice of the local-loop bandwidth. Unbundled access to loops may be required under telecommunication reform, but is unbundled access to portions of spectra required?"
*****
The upshot? Analysts at TeleChoice, Dataquest and other market-research firms who were once gung-ho on DSL warn hardware suppliers to pay attention to factors that could cool carrier interest. Dataquest's Murphy went so far as to suggest, in a Supercomm speech, that "the bridge to the 21st century will be an analog one." Even committed optimists like Amati Communications' vice president of marketing Benjamin Berry said that DSL services are unlikely this year. "ADSL rollouts clearly will be coming next year at the earliest," Berry said. "We've learned a lot in trials, but the problems of copper plant are real ones."
******
The FCC has already pressured the ILECs to stop dragging their feet on several unbundling issues, and OEMs are bringing their concerns to Washington as well. "We're currently in discussions with Clinton administration advisers on these issues," said Frank Wiener, vice president and general manager of DSL products at Paradyne Corp. (Boca Raton, Fla.). "And you can be sure we won't stand for any ILEC games with new competitors bringing DSL development to a screeching halt."
None of the problems are insurmountable. Just as DSP advances have made it possible to implement the phenomenally difficult line codes used in ADSL in just a few chips, advances in line test may make loop problems seem minor in five years. But the turmoil associated with telco deregulation-combined with the line characterization required to support new local-loop service-should assure impatient customers that broadband-access services are not just around the corner.
Meanwhile, the model for tight alliances among chip vendors, central-office OEMs and carriers is being forged early on, in the efforts to deploy PCM modems. Unlike the high-end analog modems of the past, these special systems-sometimes referred to as "56k" modems-double line rate by removing one data-conversion process from modulation.
Digital termination
Because central offices or POPs must be digitally terminated to support the fast service, proponents have gone to great lengths to educate both carriers and end users. For example, U.S. Robotics Inc. (Skokie, Ill.) launched a program for end users in early June, pledging guaranteed free upgrades to the official V.pcm standard as soon as the TIA and the International Telecommunications Union (ITU) finish their work. U.S. Robotics also offered a free month of Internet service to spur sales, and published a nationwide directory listing ISPs that support the fast modems.
Ironically, the two groups promoting competing PCM algorithms-the U.S. Robotics/Texas Instruments "X2" camp and the Open 56K Alliance, led by Rockwell Semiconductor and Lucent Microelectronics-are actually very close in their implementation details. Neither system will require major software upgrades to move to TIA and ITU compliance. But the marketing war to capture central-office equipment vendors and ISPs requires that both modem camps mount unprecedented educational campaigns up front.
"Some of the hype may seem silly right now, but the level of user awareness and of educational campaigns resulting from the 56k war benefits everyone," said Charlie Gonsalves, business and marketing manager for DSP products at Texas Instruments Inc. (Dallas)
The development of hybrid modem cards for the central office, merging BRI and analog-modem functions, makes it easier for one site to support modem pools and digital connections. But the advantage of large modem pools is greater for ISPs and CLECs than for ILECs. That's why ISPs may lobby heavily for analog modems, even as the ILECs root for DSLs.
"The ISP is the key customer buying modem functionality, then terminating calls over a T1 coming from the telco-the telco would rather not deal with so many modems," said Ray Wright, business unit director for analog service at Motorola Inc.'s Information Systems Group (Mansfield, Mass.). "But since a DSL has to terminate in a central office, the advantage the ISP holds disappears with DSLs. They could offer DSL in the future through some form of unbundling deal, but there's not a great incentive for the ISP to move to DSL."
The dynamics of the V.pcm deployment model have made ADSL proponents, in particular, rethink the notion of a residential market for very-high-bandwidth asymmetric service. Many residential customers could find 56k modem service acceptable, and reasonably priced by the ISP. In contrast, DSL's pricing structure may be more applicable for business customers-even the lower-speed and single-pair versions of DSL, such as multirate (MDSL) and symmetrical (SDSL). Dataquest cautions that typical monthly revenues realized by a carrier for a business DSL line would be around $500, compared with $37 a month for a residential DSL service. OEMs and carriers will likely converge on a model that pushes V.pcm modems for the home and DSL for business.
Other factors preparing the market for DSL are advances at both the chip and system level in supporting more complex mixes of serial digital wide-area-networking access, in products designed for both the central office and customer premises. Simple remote-access servers and T1/T3 concentrators are being replaced by complex multiservice switches.
Even at the level of the serial communication controller, semiconductor vendors are finally upgrading T1 framers and high-level data-link control (HDLC) processors for the first time in years. New HDLC devices from PMC-Sierra, Galileo Technology and other vendors are a direct response to the growing complexity of WAN access.
"People think of T1 as a mature technology, but new types of traffic are being concentrated over T1," said Steve Perna, vice president of business development at PMC-Sierra Inc. (Burnaby, British Columbia). "Users want more muxing options to fill that big gap between T1 and T3 [45 Mbits/s]." ILECs, under new pressure from CLECs, have dropped monthly T1 lease rates to around $500. They rarely sank below $1,000 a month before.
In fact, HDSL (high-bit-rate DSL)-the original DSL flavor, devised as a direct tool for carriers to provide repeaterless T1-is being reconsidered as an option to offer to end users, rather than as a provisioning enabler. Dan Cordingley, director of marketing for telecom products at Level One Communications Inc. (Sacramento, Calif.), said that a true end-user market for HDSL would probably require a move to the single-wire-pair HDSL-2 standard still in development. Nevertheless, "T1 links already are a status symbol for some work-at-home executives," he said, "and that could indicate T1 is really close to becoming a mass-market service."
ADSL's promise
The boom in promises for ubiquitous broadband access, of course, came with the advent of ADSL in the early 1990s. Pioneers like Amati and Aware Inc. developed first-generation products based on a perceived end-user interest in video on demand. ADSL designs went into a brief hiatus in 1993-94 as interest in such systems waned, but the explosion of the World Wide Web in 1995 rekindled interest in ADSL as a vehicle for Internet access rather than video.
******
Although the American National Standards Institute chose discrete multitone as the ADSL line code in 1993, continued efforts by the former AT&T Paradyne in behalf of carrierless amplitude/phase modulation (CAP) codes caused ANSI to reopen up its decision-making process late last year. Once Paradyne split from AT&T, its CAP development group became the separate corporation Globespan Technology Inc. (Middletown, N.J.), tasked with selling CAP chip sets, the de facto majority ADSL option.
*****
Motorola's struggles to sample a single-chip DMT transceiver-the long-delayed CopperGold-and the long journey by Analog Devices to trim a DMT set from seven chips to three have turned some OEMs to the non-standard but mature CAP technology. Armando Geday, the former vice president for modem technology at Rockwell who recently became Globespan's chief executive, said the boom in interest in ADSL modems by telco OEMs and startups has turned the CAP chip-set business into a commodity market ruled by the same pricing constraints found in analog modems.
To shore up DMT's position, Alcatel Telecom has unleashed its Alcatel Meitec semiconductor subsidiary (Brussels) to make an all-out assault on the U.S. market with its DMT ADSL chip set. The coalition of ILECs that specified an Alcatel multiplexer design also required Alcatel to license its chip design to other sources, and SGS-Thomson Microelectronics will be offering DMT chips based on an Alcatel Meitec design soon.
In the wake of the ADSL revival, new concepts for low-speed SDSL have emerged from Level One, Rockwell Semiconductor's Brooktree division and other vendors based on the 2B1Q line codes used for ISDN. New line codes keep coming out of the woodwork, as well, such as the "Poet" algorithms for HDSL-2, developed by Adtran Corp. and licensed to Siemens Components Inc.
One concept for 2B1Q is to offer mapped ISDN 2B + D channels in a DSL carrier, a technology Ascend Communications Inc. calls "IDSL." Felix Diaz, chief technical officer at Interphase Corp. (Dallas), said that IDSL add-in cards have an immediate appeal by mapping to a service many users are somewhat familiar with. Interphase could not see a rational near-term deployment model for ADSL to justify board-level development, he said.
Ascend's MultiDSL program seeks to transition central-office carriers from IDSL through SDSL to ADSL and higher. U.S. Robotics has a similar program called V.Everything. It plans to use a common base of 320C6x DSP processors from Texas Instruments as the core for both V.pcm analog modems and DSL line codes. Skeptics wonder whether a single board-level modem can be reconfigured for both analog access and DSL services, however. In any event, TI is also working with Amati to gain experience with DMT line codes.
A few brave developers-notably Orckit Communications Ltd. (Tel Aviv)-have experimented with very-high-bit-rate DSL, a 51-Mbit short-range copper service usually intended as a front end to fiber-to-the-curb infrastructures. Nigel Cole, vice president of business development for Orckit's U.S. operations, stresses that the company is interested in rates spanning all RADSL and VDSL rates. "The 768-kbit market opportunity appears especially important to us," he said. Line codes are not a matter of dogma with Orckit, and the company is continuing in-house efforts on CAP and quadrature-amplitude-modulation (QAM) designs, as well as DMT.
Some semiconductor vendors with unique DSP line-code talents have been yanked into DSL markets without exactly intending to go there. Pini Lozowick, director of marketing for broadband access at Broadcom Corp. (Irvine, Calif.), said the cable TV industry kept his company's engineers plenty busy, using the QAMlink chip set in designs for coaxial cable. Northern Telecom approached Broadcom in mid-1996 with an idea for using QAM in DSL applications. Lozowick said that Broadcom was at first unenthused enough to quote a relatively high price for Nortel. The Canadian telecom giant quickly met it.
"We not only got an offer we couldn't refuse, but several other OEMs started approaching us after that with QAM concepts for DSL," Lozowick said. At the recent Supercomm, Next Level Communications Inc. was showing off QAM-based DSL interfaces to its digital-loop-carrier systems.
Anyone who expected a consolidation of line codes in the name of standards may be surprised to learn that ANSI has not only created an ad hoc group for CAP ADSL standards, but later added QAM to the team's duties. In fact, mutually incompatible line codes may serve the needs of some small carriers seeking a differentiated DSL service.
The old vision promoted by the former unified AT&T-for universal service through universal standards-has been ripped asunder, said Ken Krechmer, industry modem analyst and editor of Communications Standards Review, first through divestiture and even more by the telecom-reform act.
"We're already seeing this in the X2 vs. K56flex V.pcm modem debate, but at least TIA and ITU will bring us to standards for the analog modems," Krechmer said. "For DSL line codes, the carriers are falling back to a position of keeping businesses through a proprietary advantage, rather than providing universal service through universal standards."
DSLAM arrives
The result, as Helfrich of Copper Mountain pointed out in mock horror, is that more than 60 companies have declared their interest in developing DSL equipment-before the existing copper loop has been proven capable of handling fast data. In order to concentrate the various types of DSL traffic arriving at a central office, developers have defined a class of equipment called DSLAM, or DSL-access multiplexer. The architecture of the DSLAM is not always clearly defined. A joint-procurement consortium of ILECs has released specs suggesting that DSLAMs must have a cell-conversion capability for ATM uplinks, but many vendors have been promoting simple IP-routing-only DSLAMs.
Tom Cooper, director of marketing at Advanced Telecommunications Modules Ltd. (ATML; Sunnyvale), suggests that any form of DSLAM could have a short life span. Most serious carriers want to evolve from a DSLAM to a DSL-access switch, which integrates ATM switching functions directly into a DSL multiplexing system.
"Even if DSLAMs were going to be permanent, there are too many companies with too many concepts trying to appeal to a small central-office market," Cooper said. "If you assume the DSLAM will soon become part of the switch, the DSLAM craze looks even sillier."
Co-location costly
ISPs who think they can purchase low-priced DSLAM equipment if their own gear is installed in a central office may be in for a rude awakening. "Face it, co-location is not cheap, when you take into account duplication of racks, power supplies and all the tangential equipment necessary," said Robert Bauer, associate vice president for internetworking at General DataComm Inc. (Middlebury, Conn.). "Business applications for HDSL or SDSL are fairly well defined. But a small carrier would have a hard time cost-justifying DSLAMs for residential Internet access."
Ipsilon Networks Inc. (Sunnyvale) made a special effort at Supercomm to sign up DSLAM manufacturers as licensees for its IP-switching protocols. Marketing vice president Larry Blair said many smaller carriers will use DSLs almost exclusively for data traffic. A DSLAM supporting IP switching, but without a full ATM interface, may make the most sense for them.
But executives at some companies with ATM-centric DSLAMs-including Cooper of ATML and David Sobin of Ariel Corp.'s communication systems group-believe the DSL industry will create a tower of Babel if it does not move to standard backbone interfaces. Other DSL startups, such as Diamond Lane Communications Inc. and NetSpeed Inc., are trying to stay agnostic, while making sure that ATM options will be available.
A DSL carrier business may emerge early in the next century. But there's a vast array of DSL newcomers expecting to get rich quick selling equipment and chip sets.
"Lies are being told about modem performance," Jacobsen said. "In a noisy environment, a DMT ADSL modem degrades gracefully. A CAP modem, as the signal-to-noise ratio drops below a certain level, gives you catastrophic failure, particularly in an expanded RADSL spectrum."
---
Startup Woes
- ISDN has been a non-starter across most of the U.S.
- DSL is very likely to encounter the same types of difficulties as ISDN
Copyright (c) 1997 CMP Media Inc.
You can reach this article directly:
techweb.com |
