In the Matter of the Residential	)
ISDN Service Offering of		)	Case No. PUC961696 
Bell Atlantic, New Jersey, Inc.		)


February 7, 1997


A1. My name is Fred Goldstein. I am a Senior Consultant at BBN Corporation. This testimony does not represent an official position of BBN Corporation. It is prepared on behalf of the Center for Study of Responsive Law's Consumer Project on Technology. My business adress is at BBN Corp., 50 Moulton St., Cambridge MA 02138.

I have worked in the telecommunications and data network field since 1977. My experience with ISDN dates back to 1985 when I became Digital Equipment Corporation's voting member of ANSI-accredited Technical Subcommittee T1D1 (later T1S1), which was responsible for producing North American ISDN standards. I am the author of the book "ISDN In Perspective" (Reading MA: Addison-Wesley, 1992) and have taught courses in ISDN for Northeastern University and National Technological University. I have also been a technical leader in BBN's ISDN activities as an Internet Service Provider. I hold three patents in the area of Broadband ISDN and Asynchronous Transfer Mode congestion management and switching.

Q2. What is the fundamental nature of ISDN, and does it differ from analog Plain Old Telephone Service (POTS)?

A2. ISDN represents the completion of the evolution of the telephone network from the analog past to today's digital norms. ISDN is essentially a standardized digital interface on the existing telephone network, not a separate network or a fundamentally different service. The same digital switching systems support both ISDN and POTS lines, differing mainly at the line interface and associated line-concentrating hardware. An analog line requires an analog-to-digital conversion circuit (codec) and analog-line driver circuitry ("BORSHT"), and uses loop-pulse counting or touch-tone decoders for call initiation, using protocols developed in the electromechanical era. An ISDN line requires an ISDN interface card with a Digital Subscriber Loop driver, and access to a packet handler for D-channel protocol processing.

Once the call leaves the switch, the way it is handled depends upon the bearer capability selected for the call by the caller. If a "voice" call (3.1 kHz audio or Speech, which are normally handled identically) is selected, then the call follows exactly the same path whether it originated on an ISDN or POTS line. The intervening network elements make no distinction whatever. In general, this will be over a T1 channel multiplexed onto a higher-bandwidth transmission network. If a "data" bearer is selected, then the call will be over a T1 channel multiplexed onto a higher-bandwidth transmission network. The sole distinction is that 64 kbps clear-channel data calls cannot be placed over older interoffice transmission facilities, which 56 kbps clear-channel data calls and voice calls can.

Both POTS and ISDN Basic Rate make use of the same copper one-pair local loop facilities. The majority of facilities are capable of handling either. ISDN however is typically limited to a loop range of 18,000 feet without the installation of midspan repeaters or alternative loop technology.

Q3. How should this distinction between ISDN and POTS be reflected in tariff pricing?

A3. Because ISDN is simply a digital interface onto the telephone network, it is economically rational to price ISDN based upon its cost differential vis-a-vis analog service. This cost differential is entirely in the non-traffic-sensitive (NTS) basket, so the monthly charge for ISDN service needs to be higher than the monthly charge for POTS; also, the installation procedures for ISDN are somewhat more complex, so an installation surcharge is also warranted. But there is no justifiction for pricing ISDN usage at a different price from POTS usage, or even for distinguishing between calling-plan options available on ISDN versus those available on POTS.

Even though the data bearer capability can only run at 64 kbps over a subset of trunk facilities, total demand for data bearer traffic is almost certainly a small percentage of the total capacity of these newer facilities. Therefore it does not, by itself, incur significant incremental traffic-sensitive cost above voice call levels. Most incremental growth in network capacity is likely to be 64k-clear capable.

There is room for policy discussion concerning the appropriate way to handle the exceptional cases of ISDN access, where either a mid-span repeater or foreign central office service is required. Bell Atlantic currently averages these into the overall rate, so these costs are legitimate NTS costs to be shared among all ISDN BRI subscribers. They have no traffic- sensitive cost. Recovery of their cost from usage-based charges would create an economically-inefficient cross subsidy from TS to NTS.

Q4. How does Residential ISDN impact congestion of switch and network resources?

A4. The widespread deployment of Residential ISDN would actually help alleviate switch congestion problems which Bell Atlantic implies (notably in the Reisinger testimony) are beginning to impact the network.

Bell's approach, which encourages long-duration residential data users to stick to flat-rate analog lines, is therefore not only inappropriate, but counter-productive. It appears to be based on the following flawed chain of logic:

1. Internet usage is generating additional traffic load upon the network, which is causing congestion at switches that serve large Internet Service Providers (ISPs).

2. The primary use of Residential ISDN is to access the Internet and similar data services. [BEGIN BA-NJ PROPRIETARY]X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X [END BA-NJ PROPRIETARY] While this is indeed the thrust of today's market, it may well be driven by tariff policies that make R-ISDN extremely undesirable for voice usage. Since residential POTS is always available at flat rates, measured ISDN is not cost effective for voice.

3. Residential ISDN users are by inference going to cause congestion because they are mostly Internet users.

This inference is incorrect. If a user is going to access the Internet on a regular basis using the telephone network, a more appropriate question to ask is whether the user and the network are better off if he or she uses ISDN or POTS.

From the perspective of the transmission network, both ISDN and POTS are almost identical; both use 64 kbps channels for each call. ISDN may be restricted in its use of older non-clear-channel facilities, but its faster call setup time marginally reduces the required connect time. This includes the modem-training-sequence time, often over 20 seconds, that is avoided by ISDN.

Bell Atlantic's switching facilities, on the other hand, are more capable of handling high-traffic ISDN usage than high-traffic analog usage. This is illustrated by Reisinger's testimony. He discusses congestion caused when an ISP acquired analog lines on the Myersville, MD central office. On the common 5ESS switch, an Analog Line Unit (ALU) has capacity for up to 512 lines, but only 64 time slots to serve them. This allows an average line utilization of 3.8 CCS/line, and therefore its average capacity is severely impacted by ISP lines that run close to the full 36 CCS.

An ISDN caller to an ISP would normally be making a data call, not a modem call, and therefore those calls would never terminate on an ALU. ISDN-equipped ISPs generally use Primary Rate facilities. These attach to the trunk ports of the local exchange switch, which are capable of providing essentially non-blocking service. ISDN thus diverts calls away from the ALU. Overpricing ISDN usage therefore contributes to an increase in demand for large analog modem pools, which are the worst point of congestion in the local network.

The 5ESS ISDN Line Unit (ISLU) has up to 256 time slots serving up to 512 Basic Rate lines, a much lower concentration ratio than the ALU even if counting B channels (two per BRI). Therefore the impact of heavier-than-average usage by BRI callers at the originating exchange is also less severe than the impact of heavy POTS usage.

But there is no evidence presented that ISDN callers do cause more congestion than POTS callers. [BEGIN BA-NJ PROPRIETARY] X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X [END BA-NJ PROPRIETARY] Given these modest levels of usage, which are well below the per-line capacity of a fully-loaded 5ESS ISLU, R-ISDN has not been shown to be responsible for any more congestion than POTS.

The other common ISDN switch in the BA-NJ network is the NorTel DMS-100. This is even more favorable for both high-traffic POTS and ISDN. Its older-style analog-only Line Concentration Module (LCM) has capacity for up to 640 lines, sharing up to 180 time slots (though typically equipped for fewer, based on actual requirements). An ISDN-capable LCME has up to 540 time slots, with every four ISDN BRIs sharing one dedicated time slot for D-channel traffic. With 50% analog and 50% ISDN BRI, it has 480 time slots left for its 480 lines of capacity. Unlike the 5ESS whose ISDN and analog line units are separate, the DMS-100 LCME can handle both analog and ISDN BRI.

Q5. How is Bell Atlantic's proposed R-ISDN usage pricing perceived from a subscriber perspective? Why are multiple choices not better than a "one-size fits all" price?

A5. Bell Atlantic's justification for its proposed pricing is disingenuous. Clearly the tariff is designed to discourage widespread adoption of R-ISDN. Residential subscribers overwhelmingly prefer flat-rate service to measured service, even in cases where it has been pointed out that their individual usage patterns would result in a marginally lower cost under a metered-use proposal. Some states have even legislated the availability of flat-rate residential telephone service. Yet Bell Atlantic proposes what are essentially only measured rates for R-ISDN.

The proposed tariff provides optional "callpack" usage packages. These prepaid usage packages allow a subscriber to forecast the amount of time that will be spent per future month on outgoing local calls, and to purchase this calling time at a discount from the basic measured-service rate. Such forecasting is highly unrealistic: Subscribers' usage patterns vary from month to month, and in any case are not easy to forecast. This is effectively a measured-rate tariff.

The flat rate price ($249/month) in any case is so onerous as to be unrealistic. A flat-rate tariff is based on average usage, not upon the cost of nailed service. This so-called Unlimited Plan rate is more properly labeled a Semi-Permanent Circuit (SPC) rate, as it is priced based upon its use a leased line, and assumes that its average usage will indeed be "nailed" (meaning that it will be in constant use).

POTS is based largely on a "one-size fits all" pricing approach. This is appropriate because the vast majority of its cost is non-traffic-sensitive; heavy users impose only marginally higher costs than light users. While low-usage measured service residence POTS plans exist, they have a small subscriber base. The vast majority of subscribers choose unlimited service plans, with their averaged pricing. Since ISDN's traffic-sensitive costs are aligned closely with POTS, and its average usage is not particularly high, a dissimilar rate is unjustified.

Q6. What about "nailed" users?

A6. Much is made of the so-called "nailed" users who allegedly abuse ISDN by not disconnecting their calls. This is in fact a practice encouraged by Bell Atlantic's own marketing! While the proposed R-ISDN rates makes nailed usage extremely expensive, Bell Atlantic now has a viable option for nailed ISDN which is often less costly than R- ISDN. This is the Centrex tariff. A Centrex system allows unlimited calls among its own member stations. Centrex is actively marketed to Internet Service Providers who sponsor Centrex Off-Premise Extensions (OPX) at their customers' locations. If the OPX is in the same wire-center serving area, it consumes exactly the same resources as any other nailed intra-Centrex call. If the OPX is in a different wire center, then an interoffice mileage charge is levied, and three 64-kbps channels in the transmission network are used to carry the Centrex line to the served remote location. In general, a Centrex OPX will have a lower monthly tariff rate than an Unlimited R-ISDN line. Since most Unlimited users will be connected to a single destination, putting both into the same Centrex group is often feasible.

Nailed-up calls do consume resources disproportionate to those used by other subscribers. Given the low average usage, the total incremental cost per month of the average ISDN user's local calls is considerably lower than the proposed price of Callpack 60. But even a small percentage of nailed users with 720 (1B) or 1440 (2B) hours per month of usage can throw off the average.

For this reason, it is quite reasonable for tariffs to specify a cap on usage, with overtime charges beyond that level to discourage nailing. This "threshold rate" cap must be set at a high enough level so that the preponderance of subscribers do not see the cap as a realistic impediment to their usage. Some carriers have adopted 200 hours per month for this purpose. This appears to be a good compromise value, because few users exceed 200 hours of outgoing local calls per month, other than those who nail calls. Indeed, only a small percentage are between 100 and 200 hours, but 200 hours will meet with wider market acceptance. For those who have a need for nailed calls, a higher Semi-Permanent

Circuit rate can be offered as well. It must be noted that the threshold plan works precisely because the number of options is limited, and is not priced based on an assumption that usage will be predicted or metered. It is a rate-averaged plan, akin to flat-rate POTS, and is thus most likely to be favorably perceived by potential residential subscribers.

Q7. Are Bell Atlantic's proposed contribution levels correctly computed?

A7. In computing the contribution margin to be derived from R-ISDN, Bell Atlantic assumes that all users will use precisely 90% of their package usage allotment, with zero overtime. In effect, the entire subscriber base is assumed to consist of users of 18, 54, 126, 270, 450 and 1296 hours per month. This is highly unrealistic. Even if a user were clairvoyant enough to precisely predict the following months' usage, the majority of users will have a higher cost per minute than results from a 90% fill. For instance, someone who uses 35 hours per month can pay $31 for Callpack 20 plus $9 to $18 for 15 hours of off-peak to peak usage, or $45 for Callpack 60. In either case, the per-minute usage charge and the resultant contribution margin is higher than for the allegedly typical 54-hour subscriber with Callpack 60.

A user who under-predicts usage is even worse off. For instance, a 90-hour user with Callpack 60 ($45 + 60*$.01*30 = $63; $45 + 60*$.02*30 = $81) will pay $3 to $21 more than a 90-hour user with Callpack 140 ($60). Clearly Bell's typical contribution margins are more properly computed based on a more average mix of users than the 90% case used, and should take into account the highly-contributory overtime penalty rate that some users will no doubt pay.

Q8. Does Bell Atlantic's Centrex agenda weigh in?

A8. It appears that Bell Atlantic is actually filing this unappetizing R-ISDN plan as a way to encourage higher sales of Centrex to ISPs and to organizations with significant numbers of telecommuters. The two major uses of ISDN are for Internet access and Telecommuting. In both cases, Centrex provides for unlimited use to a single destination. Residential flat-rate POTS is complemented by flat-rate Centrex, not by the proposed Callpack-priced ISDN, or by the much costlier Unlimited Plan. Bell Atlantic has the highest Centrex penetration of any major telephone company. It appears that this R-ISDN tariff is fundamentally a tool to promote Centrex, not a service intended to realistically reach broad levels of penetration.

Q9. Is ISDN really such a new, immature service?

A9. Bell Atlantic's stated perception of ISDN is at variance with reality. Koeppen states [p 25 line 7] that R- ISDN "is not today, nor will it be by the end of the decade, a service that can be characterized by anyone as a true `mass market' product...." This is apparently a self- fulfilling prophecy: If the monopoly provider prices a service to avoid mass-market levels of demand, then a mass market will not develop. In effect, Bell Atlantic is condemning New Jersey to a backward-looking era of analog telecommunications.

Furthermore, Koeppen's characterization of ISDN as "in the early adopter phase of its product life cycle" [p. 30 line 8] can only be viewed as the result of telephone company intransigence, not technology. ISDN development essentially began in 1980. The first public demonstration of ISDN switching in North America was ten years ago, the February, 1987 Phoenix trials. ISDN switches and applications were featured at the Telecom `87 trade show in Geneva, with numerous vendors of ISDN switches and terminal equipment. Indeed ISDN TE has gone through at least three generations of evolution in the past decade. ISDN Centrex has been available in parts of the United States since 1988. This is a curiously long time for an "early adopter" phase.

ISDN's acceptance in other countries has been far greater. Germany, for example, has a consumer-level mass ISDN market that greatly exceeds in size the entire United States ISDN market. This comes about in large part because Deutche Telekom has positioned ISDN favorably. It is priced at less than twice the analog-line rate, with ISDN calls (voice and data) priced at parity with analog-line calls. ISDN line ordering and provisioning is geared towards high volumes. Compatible equipment is cheap, with PC cards in the $100 range. This is not a recent development: ISDN has been widely available in Europe for several years.

ISDN's widespread acceptance is dependent on its positioning vis-a-vis POTS. A tariff that pits measured ISDN against flat-rate POTS is one that cannot succeed in positioning ISDN as more than a permanent low-volume boutique product. A cost-justified threshold-rated ISDN, on the other hand, can lead to a flourishing of the service and a dramatically higher level of acceptance.

[Return to Consumer Project on Technology Home Page ][Back to CPT Telecom Page ]