Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : Westell WSTL -- Ignore unavailable to you. Want to Upgrade?

To: Vladimir Zelener who wrote (12776)	8/20/1998 12:15:00 PM
From: Mike Harnack	Read Replies (1) \| Respond to of 21342

I couldn't help but forward the following: It clearly articulates the limitations of DMT (and consequently DMT lite) and cable/data modems. These come from non other than the father of DMT John Cioffi and Microsoft. The mass deployment issues as well as the "splitterless" issues have all been sucessfully solved by Paradyne's MVL technology. You are correct in pointing out that an end user doesn't give a hoot about which line code is the "standard". Users want service and they want it now. Mike Skeptics Confront Local Comm Services techweb.com A wave of cautionary words broke over a session on high-bandwidth services to the home at the sixth Hot Interconnects conference. In separate papers, researchers warned of unsolved problems looming on the near horizon for both Digital Subscriber Line (DSL) services and for cable modems. John Cioffi, associate professor at Stanford University, described the work of his research team on emerging challenges to DSL. Cioffi homed in on two particular issues: the complexities of mixing plain old telephone service (POTS) and DSL service without a splitter; and the problems of crosstalk. On the POTS front, "there are many issues" with splitterless DSL, said Cioffi, "including what happens during a Ring signal. But one of the most interesting is the problem of insertion loss." When a user picks up the telephone handset in a splitterless service, the result is an immediate change in load on the local loop, resulting in a loss of one to two orders of magnitude in signal amplitude, Cioffi said. "The modems are designed to recognize this, and to interrupt the data service while they retrain the modem for a lower data rate. When the phone goes back on the hook, they interrupt again and bring the rate back up. DMT [digital multitone] is designed to allow for adaptation to slowly changing line conditions, but not to abrupt changes like a handset going off-hook. That requires interrupting the data flow." One potential issue is how tolerant system software will be to these interruptions and changes in rate. Cioffi described techniques in development at his facility that would combine rate-loading and margin-loading techniques to guarantee a minimum data rate, then load for an achievable rate greater than that. Such techniques, Cioffi believed, could operate continuously, responding to even abrupt changes without interruptions. Also, Cioffi added that crosstalk can occur when several POTS twisted pairs in the same bundle are used for DSL service. "Traditionally, the industry has treated crosstalk as Gaussian noise and attempted to filter it out," Cioffi said. "Our research indicates that it is better - ultimately up to 30 dB better - to treat crosstalk as a separate signal, and to try to detect both the intended signal and the crosstalk signal." The result is vastly better usable bandwidth for the DSL. But the cost is vastly more computing power. "The computing requirement for a DSL today is about 100 Mips," Cioffi said. "Crosstalk mitigation requires up to several hundred more Mips - it could quadruple the size of the computing job. Consequently, we won't see it implemented for a year or so." But the researcher expressed confidence that more efficient algorithms would be found. While Cioffi focused on DSL issues, Microsoft's Chuck Thacker zeroed in on the unsolved problems of cable modems. "I'm cautiously optimistic about cable," Thacker began, "but I want to sound a cautionary note." He warned that the available raw bandwidth in a branch of a hybrid fiber co-ax system - 27 Mbits/s downstream and 10 Mbits/s upstream - is "peanuts." In a heavily loaded branch, he observed, the downstream bandwidth available to any one cable modem user is about the same as it would be with a 33-kbits/s modem. But the critical bandwidth problems come when the user of the cable modem is not simply surfing the Web. "What if you are a 'complex' customer, with, say, two PCs and a printer?," Thacker asked. "Cable modems don't provide routing capability, or even any way for you to talk to your own printer. Small businesses will end up requiring a pocket router to mediate between their own LAN and the cable system." And if the small businesses wanted a Web site, Thacker observed, they would almost instantly overrun the tiny upstream bandwidth allocated to them by the cable company. In addition, Thacker worried about quality and reliability issues. "What happens when the head-end goes down?" he asked. "What happens is that every modem on the system begins the 'ranging' process to establish the time delay between itself and the head-end. We have no simulation data to suggest what that will be like." Finally, Thacker observed that the cable operator would be obliged to provide a company-owned modem, upgrade his plant, increase his level of repair service and provide intensive, complex customer support out of an estimated $40 per month revenue. "As long as there are only a few subscribers, they will all see great service," Thacker predicted. "But there are only 250,000 cable modem users in the U.S. today. Experience with a wide range of other distributed systems tells us that every time you scale up the system by an order of magnitude, you encounter new problems of types you hadn't seen before. It's going to be interesting." For more technology news, visit techweb.com

To: Vladimir Zelener who wrote (12776)	8/20/1998 12:45:00 PM
From: Trey McAtee	Respond to of 21342

vladimir-- well, the big three in computers want to be able to ship ADSL modems as part of their systems. its irritating to them that the phone companies want to settle QOS issues first. it would have been better had MSFT, CPQ, and INTC gone to WSTL. this is the problem...phone and electricity are so dependable now that we all take it for granted. the ILECs dont want to corrupt that. of course this is something MSFT would never understand. they make money by releasing buggy products. good luck to all, trey

To: Vladimir Zelener who wrote (12776)	8/22/1998 7:09:00 PM
From: DX	Read Replies (1) \| Respond to of 21342

>>You might be 100% correct that the future of ADSL (assuming there is one) will be DMT, but I agree with the point Roger made, that RBOCs initially will be better off starting with field tested WSTL's CAP. << One of major problems for WSTL can be that RBOCs field-tried WSTL's CAP and opted for other technologies like DMT from other vendors. If you go to ADSL Forums, you will rarely hear anyone talking about CAP anymore. The trend is DMT_ADSL+ATM+Microsoft's software ATM SAR and SVC signalling stack. You can not simply resist the trend. Besides, DMT-ADSL modems and DSLAMs are starting to roll out of R&D labs in leading ADSL vendors like Alcatel and PairGain which virtually announce the death of CAP-ADSL. Regards,