Air traffic control designers wrestle with technology refresh
For years, leaders of the U.S. Federal Aviation Administration struggled with outmoded and obsolete air traffic control (ATC) equipment ? the famous
By J. R. Wilson
For years, leaders of the U.S. Federal Aviation Administration struggled with outmoded and obsolete air traffic control (ATC) equipment ? the famous "vacuum tube" problem. They were to correct the problem with a series of multibillion-dollar upgrades and replacements, an effort they pursued in cooperation with the U.S. Department of Defense (DOD), which had a domestic ATC system that was equally outdated.
Among the programs involved are the Standard Terminal Automation Replacement System (STARS), Automatic Dependent Surveillance-Broadcast (ADS-B), Display System Replacement (DSR), Host and Oceanic Computer System Replacement (HOCSR), Digital Airport Surveillance Radar (DASR), Operational and Supportability Implementation System (OASIS) and Air Traffic Management System Development and Integration (ATMSDI) task order contract.
In every case, a central theme was "open architecture" ? hardware and software ? to ensure the FAA and DOD never again fell victim to total obsolescence. However, as with every government program seeking the same guarantee, procurement schedules, system lifetime requirements, and the need to use commercial components quickly collided with a civilian technology base that gives most electronics a life expectancy measured in months, rather than decades.
As some of these programs, such as DSR, reach completion and others move rapidly toward installation, the question is how well have the government and contractors dealt with the open architecture issue. Is it even possible to have a 12-year procurement and 25-year operational life for commercial off-the-shelf (COTS)-based computer-dependant systems?
Given its size and complexity, perhaps the most vulnerable of all these efforts is STARS. This system combines state-of-the-art microprocessors and 2,000-by-2,000-pixel monitors on 27-inch diagonal screens. STARS can merge as many as 16 different radar images, and display as many as 900 aircraft simultaneously and six levels of weather intensity.
The open architecture effort is progressing well because the programs have become "hardware independent," reports Peter Dunham, deputy manager for domestic ATC at Raytheon Command, Control, Communication, Information Systems in Marlboro, Mass., the prime contractors on STARS and heavily involved in several other systems.
"In the old days, we had computers that were made by a single manufacturer and required a particular operating system (OS) or software, which pretty well committed everything to that computer because the software had to be tailored to the performance capabilities of the computer," Dunham says. "And they became very difficult to separate."
STARS is designed to run on the UNIX operating system, which many experts say was created to manage programs rather than interact with humans in the way a PC-based OS such as Windows does. UNIX has become "a kind of universal standard" for such hardware vendors as IBM, Sun and Digital, Dunham says, and has enabled Raytheon to run its AutoTrac system, on which STARS is based, on three or four computer hosts successfully on the same day, using the same software.
"This addresses the problem of hardware obsolescence, because as long as it can run UNIX, we can run on it," Dunham says. "And as technology advances, you can, with very little expense, port it over. On STARS, where the deployment will take several years, I suspect we will upgrade the machines to the latest versions every two years with no visible impact on the program. So the advances in computer hardware and some software ? especially given the hardware has become so cheap ? are not a serious problem.
"We're already on our second generation of hardware on STARS without any significant effects," Dunham continues. "And we already know what the next update will be. We have a specific program within STARS to manage this activity ? when do we break in new hardware and software. If you go back 10 years, with smaller memories and performance, the software really had to be tailored to the machine. Now performance is so high and memory is so cheap, standardization is easily done without a penalty."
Which is not to say all computers are the same. Dunham says each has its own idiosyncrasies ? The Alpha microprocessor has its high-order bits in the reverse order of Sun computers, for example. There are several versions of UNIX (Raytheon based theirs on the Berkeley version) and some areas where it cannot currently be used. The safety processor in the Wide Area Augmentation System (WAAS), for example, cannot use the UNIX baseline because UNIX is not certified to avionics standards.
From the FAA perspective, one of the biggest problems in updating the system is old software within and the need to continue dealing with external old, or "legacy" hardware.
"We're still in the process of redesigning the software, but still have legacy software in the system and it is very difficult to transition out of it," says John Cardina, the FAA's director of architecture and investment analysis. "The way we have mitigated that is by bringing in the new applications on local area network configurations, taking advantage of the networking technology that has come along, allowing us to develop new applications despite the legacy software. That bridges the gap somewhat."
Developing the new applications on modern computers also means dramatically changing the FAA's system refresh cycle, he acknowledges. Where in the past FAA officials could work with a six-to-eight-year cycle, that has now been cut in half, at best. And the old FAA approach of building a system or application and leaving it in place for 20 years simply can no longer be supported.
"The biggest problem we have always had is how to transition safely from an old system to a modern one, given we have many, many units in the field," Cardina says. "For some, it can take as long as seven years to transition, with safety as our paramount concern. The other factor affecting deployment of new capabilities is that the more of those we deploy, the more other systems we need ? they are no longer independent but need new datalinks to new avionics and so on. So more and more the cycle also must incorporate the avionics world, where transitions can take decades. For example, we are still supporting 50-kilohertz radios that are flying around. That makes the system more and more complex."
Open architecture has been incorporated into all FAA efforts since the agency cancelled the Advanced Automation System contract, the agency's original upgrade plan, in the mid-1990s.
"Open system technology has been a tremendous benefit to easing the development cycle and getting products out there, but technology refresh and the marketplace evolution ? which is much faster than our development cycles ? are giving us a number of problems," notes FAA chief system engineer-en route Jay Merkle.
"We're putting out large numbers of separate software packages, each with its own evolution cycle and market pressures, which gives us a rather complex life cycle problem," Merkle says. "You can't send updates to the field every day; you have to put training in place and give a facility a reasonable time to adjust to each new system. And the marketplace has been very unpredictable. We project end-of-life and end-of-service, but, because of market pressures outside the FAA, those things change, which makes it difficult to synchronize with our long-term planning and budget."
The result has been a change from proactive to reactive planning, which is one thing for the agency and quite another when dealing with Congress, whose procedures have not yet adjusted to the new realities and pace of technological change.
"Planning proactively, when you go to the budget hearings and talk about technology refresh without saying exactly what you are going to do, is a difficult story to explain," Merkle says. "Having helped develop the architecture and now being in a position of helping with the long-range planning, it has been a very good, high-level model for technology research, but we're still learning how to bridge that high level, long-term model into a near-term budget projection," Merkle says. "The architecture does work well, but when you get to the actual near-term budget, people want increased detail and fidelity. So in the two- to three-year range, as vendors change their projections, it gets difficult. I'm not sure the military, the FAA, or anyone has enough data to predict that bridge because of the difficulty of the market forces."
These same problems apply to the highly active international ATC upgrade market, but there are significant differences in system size and complexity as well as requirements outside the U.S. While the software is just as complex, no other single system has the number of units required by the FAA/DOD ATC effort. In fact, most international systems range from one to 10 individual sites, compared to more than 300 for the U.S. Even those with large geographic areas, such as Russia and China, are coming from a much more limited legacy system and modernizing (or newly installing) only a few systems at a time.
But, as with the U.S., the international community also is demanding the cost savings of COTS-based systems.
"Many customers have a specific way they want the displays presented to their controllers; sometimes we have to develop those displays in the native language of the country," says Debra Winchester, Raytheon's director of international business development for ATC programs. "For example, in our large program in Canada, all our displays are in English and French. We have a system in Samara in Russia that we implemented in Cyrillic, although the majority do have their systems in English.
"We also have to integrate our system with the external systems already existing in the client country ? radars, for example," Winchester continues. "Many customers also have additional functions and procedures they want us to modify the system to accommodate. But, for the most part, the majority of international programs are using the system we have off-the-shelf."
One concern on the international level is making certain the hardware suppliers have local support in the customer country, although the political and market changes of the 1990s have lessened that challenge, Winchester says.
"It is very rare for a customer to ask us to use a local manufacturer for computers; it's happened in the past, but now they will go with a major supplier," she says. "We do have customers who have been using a certain manufacturer and have a lot of that equipment or a special support link with that manufacturer and we are always able and willing to accommodate that."
Such accommodation is largely possible because of a layer of software that stands between the operating system and the ATC-related applications software.
"That 'middleware' gives us the ability to reconfigure all of the hardware and software in the system, take individual units in and out of service for repair or replacement, switch over to a standby in case of a failure, and perform overall monitoring and control functions. That gives us another area of (hardware) independence," Winchester says.
That independence has resulted in a largely new role for the ATC prime. For STARS, Raytheon does not produce a single piece of hardware, which is considered a commodity and provided by COTS vendors. Instead, the company writes the bulk of the software and serves as a systems integrator, a role that is increasingly common, especially in civil programs.
Nonetheless, the prime contractor remains responsible for keeping the system ? including hardware ? as current as possible throughout installation. Future technology refresh will be a separate contract, but the same problems will apply. For the U.S. STARS system, that means dealing with hundreds of sites and thousands of computers, displays, and air traffic controllers and technicians who will need to be trained ? and retrained ? for each update. While the numbers are relatively low per nation, the same applies worldwide.
And while hardware accounts for only one-seventh of the cost of deploying STARS, it will be a major cost and complexity problem for the future of the system.
"We don't talk about replacing chipsets in boxes ? it's always a complete box replacement because by the time they do any upgrade at a specific location, that box is likely to be old," Dunham says. "The STARS system is a huge investment, more than a billion dollars. The FAA can't afford to do that every couple of years. However, it will be rare that the hardware needs to be replaced except for the very oldest ones. If I can't get a replacement part, I can mix them within the system. Because there is a standard OS, it doesn't matter which box I put it in."
That is possible through the use of universally accepted hardware interfaces and avoiding all unique hardware capabilities.
"You can't use some feature that one manufacturer puts into a box that nobody else has. When we started our software a few years ago, we based it on a very well established and universal OS and did not allow our designers to use some unique aspect of any box. And that is really not a restraint because it does not force us to deliver any less of a product," Dunham says.
"What we've done is taken a lowest common denominator of a very standard tool and limited ourselves to that tool," Dunham says. "But because it is a universal tool, it is very transportable. And because we aren't limited to the specific capabilities of a certain piece of hardware or software, we can become very independent."
Maintaining that independence, to some extent, also means knowing as much as possible about each manufacturer's long-term plans. Specifically officials need to know where they are heading with technology. This is a difficult proposition, given the intensely competitive nature of the business and the fact that rapidly changing technology can quickly scuttle long term plans.
For that reason, Winchester says, they look for computers that have the ability to upgrade in place with a new microprocessor or more memory after the system has been installed.
"As part of that, we try to pick a computer where we do not need the full memory capacity or CPU capability available at that time so we can upgrade as needed later," she says. "This open systems approach gives us and our customers a foundation to keep their systems from going into obsolescence. It still requires ongoing logistics planning, but the upgrades are much easier and much more compatible and require less change in the system.
"For example, ATC has relied on ground-based systems primarily in the past, but now there is a transition to satellite-based systems," Winchester says. "The same AutoTrac system we already have delivered around the world is capable of integrating satellite-based systems, as we already have done in Mongolia."
STARS was conceived with a preplanned product improvement (P3I) program to keep it up-to-date with the latest technologies. A list of such upgrades already has begun working through the pipeline.
One such improvement is the passive final approach spacing tool (pFAST), which is software designed to help the controller get aircraft lined up more easily into final approach spacing and thus minimize wasted landing slots. pFAST integrates surveillance data, flight plan data, aircraft performance characteristics, and approach procedures within a central database. In later developments, it will become more "active" ? aFAST ? using real-time aircraft performance data to generate speed and heading advisories to refine the recommended spacing and sequence.
Another P3I element is ATMSDI, which has tools that are to go into either the en-route system or STARS. ATMSDI covers a wide range of new concept, technology and tools developments for collaborative decision making, airspace modeling and design, flight deck systems, aviation human factors, complex airspace management, airline operations center systems, and air/ground air traffic control automation. The goal is to increase runway throughput at capacity-constrained airports, improve safe operation across the boundaries of free-flight and capacity-constrained flight regions, improve the effectiveness of high density operations, lower operating costs through better routing, and increase controller productivity.
For the FAA, "open architecture" also means being able to deal with new technologies that interface with their systems, such as radio communications.
"We have spectrum problems and are running out of frequencies; our studies have concluded the best way is for the U.S. to go to VHF Digital Link (VDL) Mode 3, which allows us to get four voice channels where we only get one today," Cardina says. "By 2010, at a minimum, we want to have at least high altitude diverted to digital radio. Europe had an even greater problem and had to move more quickly, so they went to a narrow-band 8.33-kilohertz analog, which we chose not to do, going to digital instead. The advantage there is each of these four digital channels can handle either voice or data. We'll start with voice, but if we need to in the future we can go with digital. The airlines have chosen to go to multimode radios that will be able to operate in either mode."
One avionics technology that may change the very definition of ATC in the future is ADS-B, which provides the pilot and controller with the same situational awareness. As a result, FAA officials are looking at ways to share some ATC responsibilities between the tower and the cockpit. Such a change, however, would mean future upgrades to the system would have to take the aircraft and pilot into consideration rather than just the FAA equipment and personnel.
Merkle says one of the major transitions reflected in STARS has been the move from military or FAA-specific standards as the sole means of specifying systems to a stronger emphasis on services-based specifications and commercial requirements, although performance-based specifications remain a major part of the ATC arena.
"In the area of failure modes and recovery, we have requirements for fault detection and isolation and recovery that are more severe than any other users of COTS hardware," Cardina notes. "Thirty seconds recovery time might be fine for others, but for the FAA, it isn't. So while we're still using COTS hardware components, building our systems can be quite demanding. We're not out of the woods on that ? we still have to deal with it on all the systems we build. Things don't fail all that often and we're trying to build the system in such a way that we can tolerate a short outage, but it will always be a challenge to the FAA and is increasing the complexity of the system."
Nonetheless, more open standards and requirements, along with a dedicated open-architecture scheme, have helped deal with the problem of diminishing manufacturing sources that has plagued many government programs.
"Rather than seeing vendors eliminated, we are seeing far more flexibility in terms of hardware and software. And as the information technology revolution has taken hold, we have benefited greatly from the industrial standardization that has gone on independent of government requirements. That applies down to boards that do simple I/O functions. Vendors are all building to standards and giving us a broader selection of hardware," Merkle says.
"In terms of teaming with our primes, we look for a prime that will take advantage of the most available commercial technology and, where applicable, legacy systems," Merkle says. "In this sense, legacy has a very broad meaning ? it becomes legacy the minute you commission it. We want to use each system as a platform to evolve toward the future ? STARS in the terminal domain, OASIS in the flight service domain, for example. Rather than constantly reinventing, we want to look for natural technology evolutions for the functional upgrades we need to take."
FAA unveils major ATC upgrades with completion of the Display System Replacement
Officials of the U.S. Federal Aviation Administration (FAA) completed one of the two most expensive elements in their multi-billion dollar air traffic control system upgrade in mid-July with the official dedication of the final Display System Replacement (DSR) installation.
The 14 July ceremony at the Washington Air Route Traffic Control Center in Leesburg, Va., marked completion of the $1.05 billion project. DSR is replacing 20- to 30-year-old monochrome radar screens with modern color displays and modern data-processing technology at all 20 FAA en route centers. The prime contractor on DSR is Lockheed Martin Air Traffic Management in Rockville, Md.
"This new equipment improves speed, capacity, maintainability and reliability. Just as important, DSR provides a platform for future upgrades," FAA Administrator Jane Garvey said at the dedication. "Now, controllers who handle long-distance flights have modern tools that can be upgraded as needed to help deal with the strong growth in air traffic."
Garvey was citing the centerpiece goal of the agency's ATC upgrade effort: Commercial-off-the-shelf (COTS) open architecture hardware and software to prevent the kind of systems obsolescence that has plagued the FAA since the computer revolution began in the 1970s.
Lockheed Martin succeeded in bringing DSR in under budget and ahead of schedule, but officials from neither the contractor nor the FAA say they expect the road to technology parity will be an easy one in the long run. "Even if you were fully compliant with all open system standards, it's not as easy as it sounds," one FAA official says.
Fast-changing technology (and the resulting equally fast obsolescence), budgets and politics are the three major obstacles facing DSR and the other new ATC systems. While it was relatively easy to obtain congressional funding and support to replace the FAA's old vacuum tube systems, winning budget support to replace two-to-five-year old hardware or software to remain current with technology is more difficult.
Failure to do so, the FAA warns, could create the same bow wave effect as letting those vacuum tube systems remain in place too long. Congress could find itself being asked for another multi-billion dollar major system overhaul because replacement parts or software simply don't exist in 10 years ? or less.
Ideally, replacement components will be used to "technologically refresh" DSR whenever commercial elements become obsolete or can no longer be supported and maintained.
"DSR is the cornerstone in building the airspace system of the 21st century," Transportation Secretary Rodney Slater said. "DSR makes good on our commitment to the American people for the safest, most secure and efficient airspace system, capable of meeting the challenges of the new century and the new millennium."
The Washington ARTCC controls an average of 8,500 aircraft flying each day over a 240,000 square mile area covering parts of seven states and the District of Columbia. The new integrated DSR system nationwide controls an average of 123,287 aircraft daily ? 45 million each year ? flying over the largest and most congested national airspace in the world. ? J.R.W.