Harris/Dracon aims to boost its phone-repair productivity with a custom-designed, ruggedized Apple Newton MessagePad
Andy Reinhardt
Downsizing is a euphemism for the laying off of employees, but it's also a clarion call to management to work smarter and more efficiently with fewer people. The telephone companies of the U.S., transitioning from a quasi-public monopoly to fiercely competitive private entities, have had to do a lot of both in the past 10 years, and there's more to come.
The dirty little secret of labor reduction, however, is its enormous price in loss of collective knowledge. Workers who have honed their expertise, sometimes for decades, are suddenly gone, taking with them a refined understanding of problems and solutions.
The only way for companies to preserve
that legacy while reducing their employment ranks is to attempt to capture such wisdom in software, using expert systems. This need, coupled with the growing complexity of the phone system and the drive to boost productivity, is spurring a new wave of automation at local and long-distance telephone companies.
Considering how advanced the phone networks are, their operations are remarkably low-tech in certain labor-intensive areas, such as field service. Until recently, repair technicians often got their assignments over the phone from a central dispatcher (a practice that's still common at smaller independent firms) and later filled out paper forms to record the results of service calls. These forms were then manually rekeyed into mainframe computers for tracking and analysis. Even now, technicians can complete just three to four jobs per day, on average.
Phone companies are now trying to drive automation from the regional data centers into the hands of so-called craft technicians, the people in
the yellow hard hats who test and repair phone lines. There are an estimated 100,000 such professionals in the U.S. alone, and as many as half a million worldwide. Considering that they work on and around phone lines, their efforts don't currently take full advantage of the communications infrastructure that's literally at their fingertips.
Firms that supply equipment to telephone companies are now being pressed into designing for downsizing: Telephone companies are seeking products that will enable them to automate field tasks, reduce their labor costs, and improve the overall work quality. One such supplier is Harris (Melbourne, FL), whose Dracon division, located in Camarillo, California, designs and manufactures telephony test equipment. An early player in CATs (craft-access terminals), pocket PClike devices that let line workers dial into service offices for dispatch instructions, Dracon has now struck out in a new direction with a hand-held computer that is based on, of all things, the Apple New
ton MessagePad.
Dracon's ruggedized, custom-designed version of the Newton handles dispatch and other craft-access functions, and it also automates job tickets and results reporting through easy-to-use screen forms. But its biggest benefit is the intelligence that it delivers into the hands of field technicians. The
Dracon SuperTech 2000 CDA
(craft digital assistant) automates test procedures, delivers on-line help, and provides access to the repositories of data, schematics, and expertise that were formerly carried by hand or in the heads of workers--or were inaccessible in the field. How Dracon progressed from its original vision of a hand-held system to the notion of customizing the Newton is a revealing tale about developing targeted solutions.
The Dracon CDA also represents a major step in Apple's effort to reposition its maligned PDA (personal digital assistant) as a vertical-market tool. (Industry wags quip that vertical markets are where failed consumer electronics go to
die; however, the Newton's situation probably isn't so dire.) Dracon's application is a desirable, high-visibility one for Apple: It involves blue-collar workers, harsh environments, and the potential for substantial productivity payback. If Apple and Dracon make this application fly, it could be a model for other vertical-market products.
Both Apple and Dracon have the opportunity to apply the technologies of Dracon's CDA to other areas. For example, their nonexclusive arrangement allows Apple to leverage off Dracon's work in ruggedizing the Newton. This could help it sell in other demanding environments, such as oil-field exploration, mining, shop-floor control, utilities, and governmental services (e.g., emergency response, traffic and parking control, census taking, and public health).
Meanwhile, Dracon chose to use the Newton partly because of its object-oriented software environment. Although Dracon's software is now focused on telephone repair, its templates, scripts, and communications
capabilities are generalizable and reusable. Dracon hopes to eventually provide these software tools to other firms customizing the Newton for field-service applications.
Testing the Loop
Dracon is a major player in the market for line-testing and conditioning equipment in the local, or subscriber, loop, that part of the phone network located between the customer site and the CO (central office), with products such as the DATU (direct-access test unit) Loop Conditioning System. The company began the project that led to the birth of the SuperTech 2000 CDA more than three years ago, before the Newton was even announced. The project's roots go back even further--to 1987, when Dracon expanded from microprocessor-based test equipment into the market for hand-held CATs, which were pioneered by AT&T and Bellcore.
Line testing and conditioning are essential aspects of POTS (plain old telephone service) because analog circuits are so subject to quality degradation. To ensure an acc
eptable level of sound fidelity, technicians have to find failed components or line breaks and eliminate noise in the circuit.
Before the advent of such devices as the DATU, these tests were performed by two people, a technician in the field and a partner in the CO. The field technician carried a large phone-like device called a butt set (named for its ability to butt in on a line) and an assortment of analog meters for testing voltage and current. He or she would call the CO, ask a worker there to provide a test frequency or sequence of tones on the line under test, and measure the results. This iterative process was easily bottlenecked when CO workers became overburdened.
The DATU eliminates the need for a technician in the CO, because the field technician can call from the butt set directly to an interactive test system and use DTMF tones to control it. The DATU resides in the CO. Using tones, the field technician orders it to switch onto a specific line and then call back, perform diagnostic
sequences, or provide a set of frequencies.
But the process is still a bit like detective work, which is why clever humans are needed to complete the process. Most craft technicians used to have only high-school diplomas; now, because of more complex digital telephony, some companies require them to hold an associate degree or pass a test. And technicians typically spend a few weeks every year being retrained by AT&T and other phone-equipment suppliers.
When telephone companies first set out to automate field service, their first line of attack was improved dispatching. Repair workers make between $15 and $30 per hour, depending on their specialties and skill level, so one of the imperatives was to send out the right person for each job, not somebody who's improperly qualified. To improve labor deployment efficiency, telephone companies implemented expert systems in their OSS (Operational Support Systems) centers, which are regional facilities that oversee many COs. Among the information housed
at OSSes are so-called outside-plant records, which are schematics of wiring plans, maintenance histories, and databases of customer-service options (i.e., whether customers have selected services, such as call waiting, caller ID, or voice mail).
When a service problem is first reported, the OSS conducts an initial automatic test through the lines. Computers in the OSS ask loop-based equipment to access a physical circuit, and the OSS then tests such factors as voltage, amperage, resistance, and capacitance on the line. These measurements are fed into host-based expert systems that compare the findings with problem profiles and provide a basic analysis of the fault. From this data, dispatchers estimate the kind of technician that needs to be sent out; this saves money right off the bat if the problem can be solved by a junior person, and it shortens the repair time required if the problem requires a more specialized person.
The second method used by telephone companies for improving labor deplo
yment efficiency was to issue dispatches by computer instead of verbally. Field technicians were given small CATs, typically computers with built-in 1200-bps modems, that they could use to tap into a line and receive marching orders. These devices were usually proprietary, character-based terminals and relied on centralized host systems in the OSS (most of which were initially DEC PDP-11s and VAXes and, later, Unix-based minicomputers). Nowadays, CATs are more typically DOS-/x86-based hand-held PCs.
In 1987, Dracon introduced a state-of-the-art CAT built around a pair of CPUs (an 8086 and an MC 6502) and a heated, backlit LCD. But the market failed to develop, and the company dropped the product in 1989. According to
Norm Schillaci
, SuperTech 2000 product manager for Dracon and leader of engineering development for the project, CATs posed numerous architectural problems. As terminals, they were reliant on the host, which afforded little opportunity for distributing processing power. I
n fact, telephone companies preferred that CATs not be too intelligent; this kept prices down and minimized the cost of replacing any that were lost or damaged.
One advantage of this approach, of course, was that new features could be implemented centrally instead of requiring individual devices in the field to be reprogrammed. Unfortunately, CATs also provided limited information to craft workers, and users needed training to master their arcane syntax.
A confluence of forces led Dracon to its next solution, which eventually became the SuperTech 2000 system. The first factor was the soaring amount of computer power that could be built into tiny devices, which made the use of traditional terminals indefensible in price/performance terms. The second factor was anticipation of demand for client/server solutions: While telephone companies wouldn't dispose of their OSS hosts overnight, breaking the hard coupling between back-end systems and field devices was an attractive first step. Third, the down
sizing imperative and growing complexity of both the network and testing procedures dictated the need for smarter, easier-to-use field devices.
Dracon conceptualized a new hand-held system that united some of the functions of a test meter, a butt set, and a CAT and added job-ticket processing, on-line help, and built-in expertise. This vision required a general-purpose computer platform because it had to be programmable, yet it also had to meet some stringent physical characteristics (e.g., small, lightweight, battery-powered, and rugged).
The need for ease of use, which meant getting rid of all the cryptic text commands used on CATs, drove the adoption of a user interface built on forms, check boxes, pick lists, and pull-down menus. Using a smart mobile device with a GUI would let Dracon hide the legacy systems in the OSS without requiring telephone companies to scrap or retool them.
Another key early decision was to use wireless technology to connect the mobile device to other systems.
One drawback of CATs is that a physical connection to a wire must be made to download service calls. Every time craft technicians attach a CAT to, or disconnect it from, the wires, they stress the physical connection points of both the phone system and the CAT. A wireless system not only avoids this problem but also permits the exchange of data when the craft technician is driving between job sites.
Ironically, wireless technology also offers greater data bandwidth than current CATs (i.e., 56 Kbps or higher, versus 1200 bps). At the same time, Dracon decided to implement a wire-line modem in the CDA as a backup in case the wireless system ever went out of commission.
The Long Road
Field devices used to be proprietary, says Dracon's Schillaci, but a few years ago, phone companies started issuing RFPs (requests for proposals) for DOS-based systems. "It wasn't that DOS was any better; it was that they [phone companies] were afraid of being boxed in by a closed system," he say
s. "They wanted the option of running third-party programs on these devices instead of relying on what the vendor supplied. And they wanted to be able to develop their own apps."
In 1991, Dracon began looking at general-purpose platforms on which to build its combo unit. The first contenders, DOS-based pen computers from Grid Systems (which is now owned by AST), didn't fit the bill. Schillaci says that while these systems were sufficiently ruggedized, they were too heavy, didn't have a long-enough battery life, and were too expensive. Furthermore, he says, their software tools and operating environment (DOS plus the PenRight shell) weren't robust enough for the target customer base and usage profile.
Dracon also considered building its own DOS-based system from scratch but ruled that out because of high development and manufacturing costs. "We are not in the business of making general-purpose computers," Schillaci says. The company also rejected the idea of supplying an off-the-shelf (or an envi
ronmentally hardened) laptop for use in conjunction with a CAT, because this would have required technicians to carry two devices while in the field. The whole point was to roll all the necessary functions into a single system.
Looking at options other than DOS-based systems, Dracon considered, and eliminated, several of them. The Eo systems, which ran the object-oriented PenPoint operating system on AT&T's Hobbit chip, were cast out for being too big and too costly. "The Eo had lots of integration [i.e., built-in cellular capability and a rich software environment], but it was heavy and very expensive," Schillaci says. "We saw it as a real disaster, and an analyst recommended against it." Systems running General Magic's Magic CAP or Microsoft's WinPad looked like they would not be ready soon enough. (As it turned out, the Newton shipped more than a year before the Sony Magic Link did, while the first WinPad systems are yet to be delivered.)
When former Apple head John Sculley announced the New
ton, with its entry price of just $800, Dracon was impressed. "We went, 'Wow!'," Schillaci says. But equally important were Apple's image and attitude. "Lots of people were trying to figure out what a PDA was, but Apple had come the closest," he adds. "They had set up their own division. They had a reputation for being a pioneer. And they had signed up a lot of third-party partners, like Motorola and Toshiba." And, he notes, a consultant told Dracon that he expected only two pen/PDA environments to succeed over the long haul: the Newton and the WinPad.
Although the Newton was aimed at the consumer market, Dracon decided to tailor the device for a vertical application. A key factor in this decision was the object orientation of the Newton operating system, which Dracon believed would make two things possible: the reuse of code and easy third-party licensing of software components. This meant not only that Dracon could obtain software functions from other firms rather than inventing them from scratch (e.
g., the SuperTech 2000 CDA can use VT100 terminal-emulation software from Ex Machina [New York, NY]) but also that Dracon could sell the forms and procedures it created (e.g., a job-ticket template or a script for managing test procedures) to other vertical-market players. For these reasons, concluded Schillaci, "We decided that going with the Newton was the safest thing."
Customizing the Newton
Dracon's original intent was to use an off-the-shelf Newton with an external communications module connected via the LocalTalk port; the target weight for the whole package was 2 1/2 pounds, with an operating life of 8 hours per charge.
But problems emerged quickly. A wide variation in field-lighting conditions dictated that the display be backlit, which would use up more energy and require retrofitting of the unit. More important, the Newton's packaging -- especially its unprotected screen surface -- was not up to Dracon's environmental specifications. Among the problems: the degr
ee of safety from precipitation, the quality of the seals, and the placement of the speaker (see the table
"Environmental Requirements for the Dracon SuperTech 2000 CDA"
). For these reasons, Dracon decided to design a customized version of the Newton.
The SuperTech 2000 CDA is based on a standard Newton core, using the same main printed circuit board as the Gelato (i.e., MessagePad 120) release of the product. However, there are differences between the two models: Dracon's CDA has stronger plastic and a different case, is slightly thicker than the Newton, doesn't have an infrared port, and has a different silk screen over the LCD. The screen is also covered with a Mylar shield, and there is a door to cover it when the device is not in use. Also, the entire system is enclosed in a rubber boot to further protect it against breakage.
In addition, Dracon has added backlighting behind the LCD and incorporated a new "fuel-gauge" battery technology, which features a microprocessor
in each battery pack that manages different power and recharging characteristics transparently to the system. This will allow seamless migration from the nickel-metal-hydride batteries the CDA uses today to more efficient lithium-ion batteries in the future. The Dracon CDA doesn't include an AC adapter because it's intended to be used only on battery power and recharged on a supplied charge stand when not in use. Schillaci says that one reason Dracon chose both the Newton and the wireless technology offered by Digital Ocean (Lenexa, KS) was their respective power management capabilities, compared to other options.
The CDA itself is only part of the SuperTech 2000 system, which also includes communications components for linking the Newton back to the CO and OSS. Built into the hand-held unit is a wireless LAN based on Digital Ocean's spread-spectrum technology, used for talking to a communications bridge located in the technician's truck or in a satchel. This wireless LAN can also be used to interface
portable test equipment to the Newton (more on this later). The communications bridge, in turn, communicates to the service center via a wireless WAN; for now, Dracon is using analog cellular technology, but eventually it could employ CDPD (Cellular Digital Packet Data), packet radio, satellite, or some other emerging technology.
Digital Ocean's wireless LAN has a range of 800 feet and operates in the 902- to 928-MHz unlicensed frequency range. Instead of frequency hopping, it employs direct-spread radio technology, which Dracon says uses less power and allows for a potential throughput of up to 2 Mbps (although, due to software overhead in the Newton, the actual rate is now closer to LocalTalk rates of 234 Kbps).
Analyst Ken Dulaney, of the Gartner Group (Santa Clara, CA), points out that direct-spread technology is more subject to external interference than other technologies. But because the nature of Dracon's transmissions isn't very confidential and the signal range is fairly short, securit
y is not a big factor.
Dracon chose to adopt this
two-level communications architecture
instead of equipping each CDA directly with a wireless LAN (either on a PCMCIA card or built in) for several reasons. First, the wireless LAN creates a distributed communications hub surrounding each truck that is capable of supporting more than one repair technician. This helps push the work load away from service centers and into the field, reducing the data traffic going in and out of the OSS. It also breaks timing dependencies between mobile devices and host systems: The CDA doesn't have to have real-time access to the host, which permits greater flexibility on both sides of the communications link.
Second, the LAN also supports non-CDA devices. This means that makers of portable test equipment can equip their products with compatible radios and then link them locally to the CDA.
Third, the wireless LAN uses less energy than a WAN and operates at a higher data rate. To a certain
extent, it also protects Dracon's customers against the risks of evolving wireless-WAN standards: Rather than having to change every unit in the field to support a new technology, they only need to update the more easily replaced interfaces in the communications bridges. Of course, a wireless WAN card could still be installed in the Newton's PCMCIA slot by the customer.
The notion of a communications hub is central to Dracon's long-term vision for the SuperTech 2000. For the time being, the truck-based bridges (which are likely built on DOS- or Unix-based x86 computers) are mere relays that shuttle LAN traffic back to the service centers. But they will eventually become, or be attached to, distributed servers containing some of the "outside plant" information that's now housed at the OSS.
Thus, when a craft technician wants to see a wiring plan or look up a customer's service profile, the data transmission can occur locally and at faster LAN speeds instead of at slower WAN rates. Forms, templat
es, and customer information can be maintained in the servers and updated via periodic downloads from the service center. Job tickets can be distributed in batches each morning in real time and then collected when each job is completed, instead of on demand.
Using the CDA
Third-party players will determine how much of the craft technician's work is eventually integrated into the Dracon CDA. Initially, the repair technicians will still carry portable test equipment around with them, so the actual measurement of voltages and frequencies won't be done by the Newton. But over time, these functions could be brought under the CDA's umbrella or even designed into PCMCIA cards. To encourage such integration, Dracon will publish the HAPI (Harris API) specification, which outlines the roughly 50 function calls and 50 frames and templates that it has added to the Newton environment.
The closest kind of integration would make use of either the wireless LAN or the CDA's RS-232 serial p
ort. Makers of portable test equipment, such as CTC, Tektronix, and 3M, could reengineer their devices to contain a wireless LAN or serial interface and then develop Newton-based software to access, configure, and control the device. The result would allow a craft technician to receive work orders, perform tests using third-party devices, incorporate measured results into reports, and close job tickets from a single user interface. So far, no test-equipment makers have publicly committed to supporting the Dracon CDA, but the company is working closely with four major players, of which it expects at least one to attach a device via serial and another via the wireless LAN.
If this third party doesn't want its product to directly connect to the CDA, there's another possible integration option: supplying Newton-based support software. A device maker could provide instructions on how to control and configure its product; advice on how to interpret results; or even forms into which to enter measurements for
analysis, incorporation into a report, transmission to the OSS, or input to an expert system.
According to Schillaci, all the programs that are now in development for the CDA are graphical and "tappable." If any text entry is required, they use an on-screen keyboard for hunt-and-peck text entry, not the Newton's handwriting recognition. (There are also applications that use digital ink; more on this later.) Changes to applications can be downloaded to flash memory.
Certain Newton functions, such as the Intelligent Assistant, will not be available for craft-technician applications. But other built-in capabilities, such as the ability to record names and addresses, will be. In addition, data values that are created and used by CDA programs won't be accessible to normal Newton applications, so the data can be protected from corruption.
Coming to Market
The Dracon SuperTech 2000 CDA is still in development, with the first customer shipments planned for sometime between
March and May. Schillaci expected to begin beta-testing software in November 1994 and estimates the final code size will be about 500 KB. Dracon is currently working closely with Bell South, GTE, and Pacific Bell on its CDA, he adds, and is also talking with Bell Atlantic, Nynex, and other independent telephone companies.
Pricing hadn't been set by the time this went to press, but Schillaci says the cost will be value-added, meaning it will reflect the base $800 price of the Newton incremented by the value of Dracon's numerous enhancements and software capabilities. By way of comparison, Schillaci notes that less-functional CATs sell for between $2500 and $5000. "There really aren't any competitors for this [CDA] now," he adds.
The precise manufacturing details also were not firm at press time. The most likely scenario is that Apple will manufac-ture the CDA under contract to Dracon, using excess plant capacity in Taiwan and Japan, and Dracon will handle the final testing, configuration, and del
ivery out of its Camarillo, California, facility.
This relationship between Dracon and Apple, Schillaci claims, offers each party the best of both worlds. "Our expertise is ruggedized products--[preventing] what the environment, sunlight, and so on can do to products," he says. "Apple understood the limits of their design. We're going to ruggedize it, and then they or anybody else can leverage off that."
The selling proposition for the SuperTech 2000 CDA is thus threefold, Schillaci says. "We will save you money today," he says, through increases in efficiency and reduction in costs. He adds that "we will save you money tomorrow" by putting in a distributed, TCP/IP-based network infrastructure and reducing the difficulty of migrating your back-end OSS systems. And "we will even make you incremental revenue," he notes, because support for digital ink and signature capture on the CDA will enable the repair people in the field to sell services to customers and take orders from them right on the dev
ice's screen.
The only caveat is offered by the Gartner Goup's Dulaney, formerly a manager at Grid Systems, who has traveled down this road before. "The phone companies will want supplies of these devices for five years," he warns. "But doing special orders is not Apple's paradigm. That's what killed us at Grid."
Bringing Automation To The Field
The Challenge
Harris/Dracon's customers, the phone companies, wanted to:
-- Reduce the size of their field-service work force and boost
productivity above the average of three jobs per day.
-- Move their data-processing workload away from central offices
and into the field.
-- Break away from proprietary host/terminal systems and move toward an
open client/server model.
-- Improve their overall work quality through the use of expert systems.
The Response
These requirements propelled Harris/Dracon to:
-- Customize an inexpensive, mass-market hand-held system instead of
designing a niche product.
-- Build a communications system involving both local- and wide-area
wireless networks.
-- Obtain as many components as possible through licensing instead
of rolling their own.
The Outcome
-- The core Newton was wrapped in a tougher case and a protective
rubber boot, fitted with backlighting and wireless networking capability,
and loaded with software to automate phone-line tests and repair tasks.
-- Dispatch, reporting, and job-ticket clo-sure are simplified and speeded up.
-- Expert-system software makes it easier for technicians to conduct complex,
iterative tests and productively use a variety of third-party devices.
-- Distributed servers will eventually off-load schematics, network diagrams,
and customer profiles from central hosts.
Lessons Learned
This experience demonstrates that:
-- New platforms carry inherent risks: The Newton hasn't fared as well
in the market or attracted as much t
hird-party support as was
originally expected.
-- Harris/Dracon was surprised and disappointed by the paucity of
development tools for the Newton but believes the situation is improving.
Storage temperature range -40 degrees C to +66 degrees C
Operating temperature range -20 degrees C to +44 degrees C
Thermal shock resistance Unit will operate for 1
hour after ambient
temperature change
of -40 degrees C to +21 degrees C or
+21 degrees C to +66 degrees C
Water resistance Unit is splash-resistant
but not submersible
Physical shock resistance Unit can withstand
multiple drops from
up to 6 feet
illustration_link (60 Kbytes)
Key to the architecture of the SuperTech 2000 is its simultaneous use of two wireless networks: a spread-spectrum LAN and, for the time being, an analog cellular link from the truck to the operations center. Eventually, the PC in the truck will also house local data, which will reduce data traffic to and from the host.
photo_link (44 Kbytes)
The telephone companies are attempting to drive automation from regional data centers into the hands of the people wearing the yellow hard hats.
--Norm Schillaci, SuperTech 2000 product manager for Dracon and the project's engineering development leader
photo_link (22 Kbytes)
The Dracon SuperTech 2000 CDA uses Apple's Newton technology wrapped in a more rugged case and enhanced with backlighting, built-in wireless communications, and an improved battery. Apple is free to incorporate these changes into other vertical-market implementations of the Newton.
SuperTech 2000 Software
screen_link
The main screen of the suite is called the Presentation Manager. From this starting page, craft technicians can receive work orders, perform direct-access and multiline tests, and interface to third-party meters.
The Work Order Manager
screen_link
The Work Order Manager lets craft technicians receive dispatches from the ser
vice office, enter comments, and close out the job ticket.
The Direct Access Test Unit
screen_link
The Direct Access Test Unit screen is the heart of the CDA's automated test capability. Instead of having to punch the keys of a butt set, the craft technician merely selects tasks from this graphical screen, and the CDA performs the correct test sequence.
The Presentation Manager
screen_link
The Presentation Manager keeps track of every time the craft technician opens and closes a job; from this information, the CDA automatically builds a time sheet that can be uploaded to legacy accounting systems.
Andy Reinhardt is BYTE's West Coast bureau chief. You can reach him on the Internet or BIX at
are
inhardt@bix.com
The Newton Goes Vertical
