Telcommunications, ISP's, And The Internet
  by Arno Pensias 28 March 1997

home computer
An Early Computer System!
With few exceptions, analog technology dominates the so-called "loop plant" which connects some 100 million US telephone customers to the Local Exchange Company (LEC) switches which serve them. At the same time, however, the switches themselves (as well as the global telephone networks which interconnects them) employ digital technology.

In telephone parlance, "lines" (largely pairs of copper wires) connect customers to switches, while so-called "trunks" connect the switches to one another. Because the switches themselves can only deal with digits, each line connection requires analog/digital conversion. Because such "codecs" (or coder-decoders) cost a fair amount of money, switch designers arranged things so as to share a certain number of codecs among a larger number of lines. And, since telephone lines typically get used for only about thirty minutes a day, planners settled upon an architecture which provided one codec for every eight lines.

Such a sharing arrangement fits well with the concept of Universal Service, i.e. the notion that everyone benefits from extending telephone service to as many people as possible. Small wonder then that regulators State Public Utilities Commissions (PUC's) and the Federal Communications Commission (FCC) incent LEC's to build facilities with shared rather than dedicated resources.

And sharing sufficed, until the advent of the Internet. As more and more net-surfers began to stay on-line for hours at a time, the sharing arrangement's underlying assumptions no longer applied. Customers who were unlucky enough to get bundled in with a bunch of modem users, found themselves waiting for dial-tone. Naturally, they complained to their Local Exchange Companies. And the LEC's, in turn, added grievances of their own. While the modem users buy extra lines in large numbers, they use a disproportionate chunk of shared resources and, even worse from the Telco's point of view, seem poised to make "free" long distance calls over the Internet.

As the number of on-line households grows, so too does their impact upon US telephone service. According to a Yankee Group estimate, this country's on-line households will grow from today's some fifteen million, to over forty million by the end of this decade. Furthermore, the same study predicts that the overwhelming majority will connect via telephone lines. Therefore, even though other data access technologies such as cable, satellites, and ADSL will grow in the coming years, modem traffic will dominate telephone and on-line economics for some time to come.

In order to promote Universal Service, PUC's normally set the rates which LEC's may charge for basic telephone service somewhat below the amount they consider reasonable for sustaining a healthy business. The regulators expect the LEC's to make up the difference from FCC-mandated access charges from long distance service providers, (as well as from enhanced services such as call waiting etc.)

Unlike long distance phone companies, the FCC has explicitly excluded on-line service providers from paying access charges. In doing so, the FCC appears to have been subsidizing the Internet vis a vis local and long distance telephone service users.

At first glance, the LEC case seems persuasive. Internet users...or the Internet Service Providers (ISP's) that serve them--ought to pay their "fair share" , the LEC's maintain. ISP's, on the other hand, point to the extra LEC revenue they have generated, and argue that it is up the LEC's to figure out how to make money in today's competitive market, without raising their rates. Who's right? The answer lies in the details of the costs involved. Since the costs at issue stem from the use of facilities, we need to examine the facilities involved, how they are presently used, as well as how they might be used in the new world of (at least partial) de-regulation. Among other items, the FCC is in the process of re-writing the rules for access charges, while the Telecommunications Deregulation Act obliges the LEC's to "unbundle" their loops i. e., lease them at wholesale prices to competitors for resale to consumers at retail.

Today's Telephone Technology


While the offerings of various manufacturers differ in their details, they embody similar design principles. In a typical system, the thousands of lines served by a particular switch presently get grouped into bundles of 512. (For technical reasons, most group sizes correspond to some power of 2--such as 8, 64 and 512.) Each of these bundles, in turn connects to a so-called "line concentrator." With 512 connections on its line-facing side, but only 64 connections on its switch-facing side, the concentrator acts as a kind of funnel. When it detects a lifted telephone receiver, for instance, it connects the line in question to the switch. Conversely, when the switch signals the arrival of an incoming call, the concentrator completes the required connection in the opposite direction.

Now here's the rub. Suppose you're "lucky" enough to live in a technically-minded neighborhood. If just ten percent of your neighbors decide to go on line at the same time, they would occupy fifty-one of your group's sixty-four possible connections to the switch, leaving just thirteen free connections for all the incoming and outgoing telephone calls made to and from the other four hundred and sixty-one lines. Even worse, suppose that one of your neighbors decides to become an Internet Service Provider, begins installing modems in his garage, and orders extra telephone lines for them. Woe to the neighborhood in that case. Remember that each incoming call takes up one of the shared slots on the switch. Since most of an ISP's modems tend to be busy during the hours of highest demand, a local ISP can exert an even larger impact on shared resources. Small wonder then, that California's Silicon Valley ranks high in congestion-related service complaints.

Interestingly, field data indicates that most complaints of dial-tone delay stem from lines connected to ISP's, rather than your neighbors' work-from-home traffic. Why? Because of present-day economics. Roughly speaking, ISP's currently provide something like ten modems for every one hundred customers they have. Even at that level, something over half their service revenue goes to pay for facilities. While this level of concentration approximates the one used in LEC switches, a typical computer user generates about four times as much connection time as does a conventional telephone customer. That's why would-be Web-surfers frequently need multiple attempts to get through to their on-line service providers (and frequently shop around in the hope of finding one with more-available access ports). From an economic standpoint, therefore, the present situation finds both the ISP's and the LEC's between a rock and a hard place.

A way out?


What alternatives exist to levying extra charges for Internet-related services, and thereby discouraging its use? Fortunately, telephone technology has benefited from many of the same advances which underlie recent progress on the computer side. In particular, the advent of powerful, and relatively-inexpensive, digital signal processors allows a single chip to accomplish much of the work now done by several dozen codecs. Good news for congestion-threatened telephone users. In the coming months, every major manufacturer of telephone switches will offer one-codec-per-line add-ons to their existing products. Most do so already.

As LEC's increase the capacity of their switches by buying additional channel units, the new non-blocking technology offers them an opportunity to regroup lines with a history of high usage (mostly modems, but more than a few teen-agers might qualify as well) to the new equipment. Everyone benefits. The LEC's get rid of an annoying nuisance. Modem users can surf to their heart's content, and traditional telephone users can continue as before. Moreover, everyone will see better service. Sounds too good to be true? Let's dig a bit deeper and see.

Cost considerations


The cost of a connection from home to switch represents a substantial Telco investment. Individual costs can vary all over the lot. Adding a line from a multi-wire cable containing unassigned pairs costs little. Digging up the street to lay a new cable when an existing one fills up costs a great deal. But, since this capital cost gets incurred no matter how much a particular customer uses his or her line, it washes out in any usage-based comparison.

As for the switch itself, its costs tend to resemble the razor business. The "blades" in this case being the per-line connection equipment. Current costs vary (depending on things like the size of the purchase, and the functionality required), while historical costs lie on a fairly-steep downward curve much like other kinds of electronic equipment. Traditionally, telephone industry experience has shown the cost of switching to be a modest fraction of the cost of running wires from homes to switching offices at least as long as switch concentrators can accommodate traffic demands. Therefore, shifting as many high-usage customers as possible would help alleviate the problem at least as far as those customers were concerned.

Three caveats, however. First, even though the per-line price of non-blocking switching equipment looks to be about the same as its concentrator-based counterparts, it costs money to rewire a customer connection from one switch location to another. Second, except for a modest annual need for replacement, the demand for additional telephone lines will determine the pace at which non-blocking technology enters the telephone network. Finally, the "trunks" which connect one switch to another are entirely usage-sensitive, with either the new or the existing switching technology.

To tie these threads together, let's assume that the right-hand machine belongs to an ISP, and makes full-time use (or has a high probability of use at peak hours, which amounts to the same thing) of its switch connection. If that connection involves a concentrator, then the ISP could take up as much as eight times as much switch capacity (and its associated cost) as an average user. As the LEC's move ISP's and other "full-time" users to the new non-blocking technology, however, each one would employ about the same amount of line unit capital cost as average users above (plus a bit more for a disproportionately-greater share of the heretofore-neglected switching fabric), the cost of occupying trunking facilities, and the (one-time) cost of moving the lines involved from one kind of codec to another.

Who pays?


Networking As businesses, ISP's pay more for their lines than do residential customers, so the extra money might well defray the reduced (but still non-zero) cost of extra capital required in the new system configuration. But that won't solve the entire problem. As we have already noted, LEC's and their regulators, expect "normal" telephone lines to generate a certain amount of long-distance traffic. Under this government-mandated system of access charges, long-distance telephone companies pay local-service providers a few cents per minute on each call. Just pennies, but the amount adds up to a substantial portion of the local LEC's earnings stream... one which figures significantly in the rate-setting process.

Even with the new technology, therefore, business as usual can't satisfy everyone. What to do? ISP's and their allies don't want the Internet "stifled." If LEC's raise their prices to "ordinary" telephone users, then folks who don't want (or can't afford) networked computers will foot the bill. On the other hand, if prices were to stay the same, thereby shifting the cost to LEC investors, the traditional public telephone network would move that much closer to becoming a money-losing backwater.

According to publicly-available information, ISP's spend something like half of their famed $19.95 per month for communication services. Moreover, the majority of that half for goes for telephone lines, some for modem maintenance, and rest for actual Internet services. While America On-Line and some others, get Internet connections through their own subsidiaries, fast-growing ISP's such as EarthLink, and MSN find advantage in buying access from service providers like UUNET, so as to concentrate their internal resources on differentiable activities like customer care.

What would happen if one or more of the LEC's were to become an active service provider to ISP's? Instead of feeling themselves treated as costly nuisances, ISP's could benefit from Telco economies. Unlike the fragmented and geographically- dispersed ISP community, the Telco could place the electronic equivalent of large banks of modems on the trunk side of its switches. Since no individual ISP can afford to locate modems at every LEC switch, their connections requires extensive and expensive trunking facilities. With a shared facility, on the other hand, modem banks could communicate with ISP facilities over cost-efficient packet networks, thereby reducing the overheads associated with sending data over trunk facilities designed for voice telephony. In addition, maintenance would become simpler, and requirements for costly extras...like the number of spares required would diminish.

We could go deeper, but the point seems evident. Properly employed, technology offers us a way out of the unpleasant bind in which LEC's, ISP's and customers now find themselves. Will the above-cited savings yield enough per-line profit to make up for lost revenue from long distance access charges? While only careful analysis can answer that that question, a rough estimate suggests it not be dismissed out of hand.

Say, for the sake of argument, that the owner of a switch were to offer facilities to ISP's at one half the present per-modem rate (now something over $100 according to most analysts). Such a rate might then enable ISP's to cut their customer to modem ratios from the present ten or so, to a much smaller number. As service improves, and time on the Net increases, so should revenue from advertising, gaming, commerce and the like. From the LEC perspective, driving down the cost of serving the ISP could serve as a source of efficiency-driven profits, hopefully large enough to replace whatever long-distance access charges might bring from an equivalent number of telephone lines.

One last number. At least one manufacturer of a box which, according to its brochure, appears able to perform the "circuit to packet converter" function for a capital cost of about $200 per (virtual) modem. Whatever the actual purchase price, we can surmise that the total cost of providing the service will hinge mostly upon how one locates this unit in the network, together with the costs associated with operation, administration and maintenance.

In the final analysis, however, the actual choice of architectures may well have at least as much to do with regulation as with economics. Present FCC regulations appear to require LEC's to either serve ISP's as if they were telephone customers (with all the extra paraphernalia that such treatment implies) or handle them through a separate business entity. While not altogether hopeless as far as matters stand today, the first alternative appears to offer little room for future enhancement. The second alternative, on the other hand, incents LEC's to place an extra layer of switching between residential customers and the telephone switches which presently serve them, so as to detour Internet-bound traffic out of the public network before it can cause accounting difficulties. Naturally, the cost savings and related revenue cited above would also detour out of the public network at the same time.

Looking to the future, an increasing number of households will opt for broadband digital service straight to their doorsteps in the coming years. Someday, everyone else will also get that same kind of service. But how can we serve "everyone else" in the meantime? With so many interested parties holding large stakes in the answer, the question seems certain to provoke lively discussion.



BACK DOOR!Back To Home Page Back Up To The Top

Web Page Design Is Copyrighted ©2000 Through By Barry Rudolph