In part 1 of this series of articles, I described my experience attempting to purchase fast Internet access. In spite of claims that 56 kbps modem services, ISDN, DSL and cable modems are all available in our area, the reality is that none of them are available for most consumers. The marketing hype surrounding fast Internet service is way beyond the reality. In Part 2 I describe deficiencies in ADSL and cable modem technologies – and note that neither technology may work for significant numbers of “served households”. In particular, according to BellCore a minimum of 24% of all phone lines are not capable of supplying ADSL.

In Part 3 of this series, I’ll describe my experience with setting up DirecPC 2.0, a system that delivers Internet data via satellite.  There will likely be a Part 4 discussing other wireless solutions for high-speed access.
 

The Limitations of ADSL

The reality, though, is that far fewer homes and businesses will actually be capable of getting ADSL service than claimed in the marketing hype surrounding the rollout. For example, I live within a DSL-equipped exchange operated by U.S. West and U.S. West undoubtedly counts our home as a “DSL served home.” However, U.S. West admits to me that it not only cannot serve our home, it currently has “no strategy” (their words) for deploying to our neighborhood. In a January 18th article [5], U.S. West admits, “DSL service is available to 30% to 40% of U.S. West customers in Spokane”. In PacBell’s situation, it is quite possible that 25% or more of the 9.5 million residences and businesses served will not be able to get any service at all.

ADSL technology basically modulates a RF signal onto the twisted copper lines. To do this, the phone company (or third party) installs an ADSL “modem” in a rack of equipment at the local switch serving your area. This modem places the RF carrier onto the twisted copper lines that run to your home. At your home, you use a matching ADSL modem to decode the signal into a data stream that connects to your computer. 

With ADSL technology, a single phone line continues to carry your analog voice signal while simultaneously delivering the data signal. The voice signal is modulated onto the wire at baseband (low frequencies) while the data  signal is placed above about 200 khz in frequency. A low pass/high pass filter combination is used to separate the voice and data signals from one another at the switch and at your home. The phone companies use two kinds of ADSL: (1) splitter-based and (2) splitter-less. Splitter technology requires a phone company service technician to install a splitter filter at the point where the phone line enters the residence, and to use an existing interior line, or to install a new line, for the data signal. Splitter-less technology routes the combined signal throughout the home’s wiring and small filters are installed by the customer to the phone outlets in each room. One user of splitter-less technology in Seattle tells us that the U.S. West splitter-less ADSL creates a distinctive “background hiss” on his regular telephones in his house.

The problem with ADSL is that not all phone lines are physically capable of carrying the ADSL signal. There are many limitations regarding which copper lines can carry the signal.
 

• Many phone lines have “loading coils” placed on the circuit. These 88 mH inductors were added to improve voice signals over long lines – but they attenuate the frequencies needed by ADSL. A phone line containing loading coils is incapable of carrying ADSL signals. According to Bellcore[1], an astonishing 24%[1][2] of all phone lines have loading coils and cannot be used for ADSL service. Loading coils also, generally, prevent 56k modems from operating on the line.
• Overall line length and the wire gauge (#24 AWG versus #26 AWG) impact the ADSL technology. Generally, ADSL can be provided only to destinations within 15,000 feet of the central office switch (the actual distance depends on which wire gauge was used in the wiring). 40.4% of all wiring is 26 AWG, which shortens the usable distance.
• In some cases (like my neighborhood), service was extended via a fiber optic cable to a termination point in the neighborhood where the light signals are translated back to electrical signals on copper wires. Ironically, “fiber” in this case, means that we cannot be served by ADSL.
• “Bridge taps” also impact service availability. 56% of all lines have bridge taps, and in 5% of these[1], ADSL service will probably not work.
• Finally, the phone wire enters the private residence. The quality of inside wiring varies from good to horrible. There is no telling what will happen to the signal once it gets inside the home’s telephone wiring.

Because of these and other limitations, ADSL service cannot be provided to all customers in an ADSL configured telephone exchange. There is no way to accurately estimate the number of lines on which ADSL will work – the only way to determine service availability is to actually test the line.

What all this means is that at least 24%, and perhaps rising to around 35% of all phone lines cannot be served by ADSL technology. (Or, in Spokane it may be as high as 70% of all phone lines cannot be served). For PacBell, using the conservative estimate that 25% of the lines cannot handle ADSL service, this means that their much publicized ADSL rollout to 70% of the switches in the State will reach only about half of the homes in California.

When phone companies claim to be serving millions of customers with ADSL rollouts, they are deliberately misleading the public. Perhaps after replacing vast portions of their network, they will have the capability to serve most customers, but those upgrades will take many, many years to implement – and only if the telco decides to upgrade the physical plant. 

As it stands today, ADSL is a potential solution for some telephone customers. However, without major telephone plant upgrades, ADSL cannot serve all customers.
 
 

Cable Modems

Most systems use “Hybrid Fiber Coax” (HFC) to deliver the data stream over optical fiber direct to a neighborhood, and then use the existing TV coax delivery system to carry the data signal the remaining distance. The local coax segment may serve anywhere from about 200 to 2000 homes and acts as one great big shared and unprotected Ethernet connection.

Early cable modem systems use “one-way” technology. That means they are capable of blasting data outbound only. The return path is handled through the existing telephone network and a dial up modem connection. This technology has the advantage that it can be deployed quickly, without upgrading the one-way cable TV delivery system to a two-way distribution network. The costs of upgrading to two-way are large, and revising the infrastructure is time consuming. Therefore, “one-way” technology provides a lower cost, quick to market solution.

Cable modem technology, however, has its own problems.

• First, the technology relies on a shared network segment in the neighborhood. This means that any marginally qualified technical person can easily “sniff” the network and watch the data go by, as there is no security. All it takes is your personal computer and commercially available network monitoring software (used by software developers and network administrators to debug software and network systems).
• The shared segment is easy to jam, either intentionally or unintentionally. The San Jose Mercury News ran a story in December describing the problems encountered by @Home in their TCI-based Fremont, California network. There, an errant system jammed the cable network reducing typical user access speeds to about the same as a telephone modem. It took @Home weeks before isolating the problem.
• The return path in 2-way systems is operated at 5 to 40 MHz. This frequency bandwidth is subjected to tremendous atmospheric noise, machine-generated and other electrical interference, and of course, radio transmissions operated by Amateur radio operators, CB radio operators, and others. As such, the return path is highly susceptible to interference.
• TCI’s @Home network (at least in the Fremont system, according to the S.J. Mercury News) comes with a host of restrictions on usage. First, you are permitted to use one and only one PC with the system. If you have networked a couple of PCs in your home, only one can be used. They claim that they can detect your usage of a proxy server and will prevent you from network use if you violate this restriction. How can they detect that? Believe it or not, their systems open up your packets and check!  Further @Home is required to disconnect connections to streaming video servers when users remain connected for more than ten minutes. Their contract with TCI makes clear that video is to be distributed over the cable TV network, not the data network. Users that exceed ten minutes of streaming video will find their connection cut off[7].
• Cable modem installation is labor intensive. It requires sending one or more technicians to your home to run wiring to the location of your computer and to open up your system and install a 10-BaseT network card to interface to the modem. When you consider the time and labor needed to merely upgrade the network to be “data capable”, it is clear that the labor intensive installation process means that signing up individual consumers will be a very time consuming affair. Further, there is a significant shortage of qualified installation personnel (see "Help Wanted" and "Hiring Fever", both  San Jose Mercury News articles, and
• "@Home Rollouts Delayed in California", at Cnet news)
• Performance claims made by some cable companies are exceedingly and deliberately misleading. In their advertising, they often claim Internet access speeds of hundreds of times faster than an ordinary telephone modem. This figure is determined by dividing the network speed (e.g. 28 Mbps) by a telephone modem’s speed (28.8 kbps). However, this claimed performance ignores that the cable bandwidth is shared by potentially hundreds, if not thousands, of other users. Further, the average data delivery speed over the Internet itself is often in the 100 kbps to 200 kbps range (some web sites do deliver faster service). These limitations are due to the speed of web servers, Internet congestion, and the route that the packets take through the network. Broadband networks are implementing strategies to store copies of popular Internet web sites within the private network[7] so that broadband customers will fetch some Internet sites from a copy maintained locally. This technique will speed up access to popular web sites and content  –but  definitely not to all web sites.

Why These Companies Are Not Telling the Truth

Besides the practical observation that most technological endeavors are more daunting than first envisioned, and typically take much longer than first predicted, there are genuine technical and practical limitations in their rollouts. The facts are that significant percentages of allegedly served customers will have no service what so ever.

So why are these companies hyping their service offerings? Perhaps they are merely naïve, or perhaps this is the usual shady business practice of over-selling and under delivering.  Most customers are excited to have service that is merely faster than their landline modem and may not notice that performance does not live up to claims. Then again, these companies may have important political considerations for exaggerating.

Telephone companies have been ordered to open up their networks and to lease bare copper and other facilities to third parties, especially to Competitive Local Exchange Carriers (CLECs) so that CLECs can provide local phone service. The government ordered this because telephone companies originally operated under a government grant of monopoly. As such, telephone companies operated without competition. Currently, many telephone companies have erected hurdles or used the courts to stifle the leasing of their facilities to third parties. Regardless, there is a great demand by Internet Service Providers to gain access to the copper infrastructure to deploy their own ADSL services. Simultaneously, regulators and politicians are becoming increasingly vocal about their frustration at the slow pace of high-speed data service rollouts. 

Cable companies are not currently required to open their networks. And they are fighting to keep their networks private and to not lease space to others. Because AT&T is acquiring TCI cable, America Online and others have requested that TCI be forced to lease capacity to third parties. AOL and other ISPs suggest that they will be forced out of business if they cannot gain access to the cable bandwidth. Many cities are telling TCI that if it merges with AT&T, then TCI must open up its cable lines to the local Internet service providers.

Therefore, there is a tug-of-war underway between telcos, cablecos, Internet Service Providers and regulators and politicians. If regulators and politicians believe that high speed data services are not being deployed rapidly, they are more likely to show sympathy to the ISP requests to open up the networks in the hopes that competitive pressure will bring solutions to market, quickly. 

On the other hand, if the telcos and cablecos convince regulators that high speed data services are available and are deploying rapidly, the perceived need to force open the private networks so that competitors will bring services to market is lessened. Therefore, it is in the self-interests of the telephone companies and the cable companies to exaggerate their high-speed deployment plans. 

By presenting a mirage of high-speed services, regulatory pressures may be lessened on the telcos and cablecos. Unfortunately, this approach of relying on the telephone companies and the cable companies guarantees that many, if not most customers, will not see high-speed services for years, if ever. Realisitcally, under existing laws, telephone companies have little incentive to upgrade their networks when the law requires them to immediately turn around and lease their facilities to competitors at a government mandated discount. The same situation may happen to cable companies forced to lease their capacity to others. The result is that telcos won't upgrade bad copper wiring - or fiber concentrators - to provide people like me with ADSL or ISDN service. In effect, consumer protection legislation may discourage necessary investment - and prevent 1/4th to 1/3rd of consumers from getting high-speed access.

Rather than force open the existing structures, I am now thinking it would be better for third parties to exploit the weaknesses in the telco and cableco systems and rollout competing wireless solutions using MMDS, LMDS, digital TV “subcarriers”, unlicensed networks like Ricochet, third-generation digital cellular/PCS and satellite-delivered services. It is a fact that no single technology can provide service to 100% of the customers in a given area. Rather than institutionalize the ADSL and cable modem solutions that guarantee a lack of service to huge numbers of customers, third parties could be building out competitive wireless solutions. These alternatives would be great for customers who would see the availability of service and lower prices created by the competitive marketplace. In particular, MMDS technology has already been proven in field trials and actual service offerings (see http://www.wantweb.net) and is available now. Due to financial difficulties in the MMDS industry, these businesses would welcome an infusion of investment from ISPs – indeed, many MMDS providers could be bought at discounts. Large ISPs like AOL could easily afford to invest and partner with alternative providers and help to create a conmpetitive market for consumer access.

In summary, ADSL and cable modems cannot serve all customers. Regulatory plans to force TCI to open their cable systems to competitors, and for telephone companies to lease to third party ISPs guarantee there will be less competition and that service will not be available to vast segments of the population. I’ll venture that it will be many years before 50% of the homes in the U.S. can actually subscribe to ADSL or cable modem service. And for those who can subscribe, service for many will be provided by a single source supplier – the telco or the cable company, but not both - in other words, a monopoly..

Under likely scenarios, it looks like cable modem technology may have the upper hand, warts and all, due to ADSL’s inability to serve many customers, even where deployed. Two-way cable service, however, requires substantial infrastructure improvements and labor intensive installations. Cable companies have the incentive to make these improvements since they can then offer many new revenue producing services, including high speed data, voice telephone services, long distance phone services, enhanced pay-per-view capabilities and even retail sales. But, cable companies have one huge public relations problem: that is, their customers do not like cable companies. According to a survey by Park Associates[8],

• 48% prefer to receive service from a phone company via ADSL
• 26% prefer to receive service from a cable company
• 29% prefer a telco provider even if the cost is $10 to $20 more per month

Don't misunderstand my comments -  many consumers will get decent ADSL or cable modem high speed data services. My point is that a very large number of consumers will not get any service and marketing hype claiming otherwise is exaggeration and false promises. Existing regulatory approaches, designed to protect consumers, seem likely to slow deployment by telcos and cablecos who see little interest in investing in infrastructure only to turn around and lease it to competitors at a discount. That means less competition, and fewer served households.

Cable and telephone companies are faced with large infrastructure improvement programs to truly offer service to all potential customers. Meanwhile, there are alternatives preparing for deployment – some think that 3rd generation digital cellular technology will be able to provide high speed Internet services (256 kbps or faster) using wireless technology overlaid with existing PCS and cellular systems. Metricom plans to eventually rollout a "turbo" Ricochet network delivering 128+ kbps performance. Wireless technology must be a component of providing high speed data services in nearly all communities, otherwise, the "world wide wait" will take a little longer.

That’s my opinion,

Ed Mitchell, KF7VY
http://hamradio-online.com
 

Coming:

Part 4: Wireless solutions - and two wired solutions you might not have expected - to delivering high speed Internet access
 
 

References and Footnotes

[1] “xDSL Testing Tutorial”, published by the International Engineering Consortium, and available online at http://www.webproforum.com/tautron/topic01.html

[2] “Data Rides High on High-Speed Remote Access”, Sudhir Dixit, Nokia Research Center, IEEE Communications Magazine, January 1999, pages 130-141. This article reports that 20% of lines cannot accommodate ADSL due to loading coils.  No reason is given for the difference with the Bellcore 24% figure.

[3] “Overview of the Cable Modem”, online course notes for the WPI EE535 Telecommunications and Transmission Technologies Course, http://bugs.wpi.edu:8080/EE535/hwk96/hwk4cd96/bcbeau/node1.html

[4] The actual speed provided by ADSL depends upon several factors including: (1) higher speeds are usable only over short distances (as little as 1,800 feet) – lower speeds can travel greater distances; (2) telephone companies may choose to offer much slower speeds than are physically possible, based on business considerations.  PacBell has announced 1.5 Mbps service using a technology that  may work out to 18,000 feet on #24 gauge lines but only 5,500 feet on #26 gauge wiring. U.S. West offers a 256 kbps service in both directions, perhaps because this is all that the U.S. West lines are capable of providing. See [1], Section 5, Table 1 for information on distance.

[5] January 18, 1999, Spokesman-Review newspaper, Spokane, Wa, page A13, “Pedal to the Internet”

[6] PacBell announces ADSL services

[7] As of January 15, 1999, @Home and RealNetworks have announced a partnership to use RealNetworks technology to locally cache streaming video content within the @Home network. This means that media requiring lots of Internet bandwidth will actually be copied and stored locally within the @Home broadband network. This should improve performance and reduce the demand on the overall network. Local caching is a key component of broadband networks in order to reduce the capacity of the Internet connections and the overall demand on the Internet.

[8] Park Associates survey, http://www.parksassociates.com/press/press_surfing1.html