Ian Sheppard/LONDON
Efforts are under way in the USA to enable computer systems to a smooth transition through the "year 2000" software barrier. Defence and air-traffic-control (ATC) infrastructures have been afforded top priority, but there is some concern that, outside the USA, a dangerous "wait-and-see" attitude prevails.
The problem is relatively simple, but the solutions may not be. The numerical turn of the millennium may cause computers to crash as system clocks reset the year from 99 to 00. Even if the clocks supply the year in four digits, millions of lines of "legacy" software code in use around the globe, and written in the 1960s, 1970s and 1980s in numerous programming languages, will not recognise it. Neither will they register that 2000 is a leap year. Estimates put the worldwide repair bill at up to $600 billion.
If that is not enough, for in aviation there is an additional problem. The global-positioning system (GPS) is vulnerable to another time-related computer glitch which occurs 132 days before the turn of the millennium.
GPS time has been counted in weeks elapsed since midnight on 5-6 January, 1980, with counting carried out in units to 1,024. The end of week (or "Z-count") roll-over at the turn of the century occurs when this system clock moves from 1,023 to 0 weeks at midnight Universal Co-ordinated Time (also known as Greenwich Mean Time or "zulu" time) on 21-22 August, 1999. At that instant, any unprepared GPS receivers will revert to 6 January, 1980, creating a problem, as the GPSsystem relies on signal time differences to calculate distances from satellites.
The GPS represents the world's most significant system to be year 2000 non-compliant but, as its proprietor, the US Department of Defense, plans for all US military aircraft to use the GPS for navigation by 2000, it is evidently committed to maintaining a continuous service.
Growing dependence on GPS-guided smart bombs has heightened Pentagon concerns that integrity should be maintained, while the GPS has grown in use beyond its designers' wildest expectations, with thousands of users now operating outside the US military arena for which the system was specified originally.
The NAVSTAR GPS Joint Program Office (JPO) has indicated that the GPS space- and ground-control segments will be year 2000-compliant, as well as being transparent to the end-of-week roll-over, by mid-1999.
Consultant Peter de Jager, however, delivered testimony to the Science Committee of the US House of Representatives in 1996, in which he pointed out that, while software project-timescales are notoriously unreliable and prone to slippage, the year 2000 cannot be delayed. Schedule delays in worldwide GPS ground-segment modernisation have already forced the JPO to put $7.6 million aside to make some of the "legacy" code in the present system year 2000-tolerant - just in case.
EFFECTS ON INDUSTRY
In the space sector, some of the original satellites are equipped with atomic clocks which are not compliant with the Z-count roll-over, but these are being replaced as the satellites themselves are replaced with new ones.
In the user-equipment segment of the industry, many military weapons, guidance systems, radars and communications processors rely on poorly written code in obscure languages, further escalating reprogramming costs for year 2000 compliance. The UK Ministry of Defence has estimated that it will spend around £100 million ($167 million), but says that this figure is likely to be revised upwards.
In the commercial sector, most equipment is newer and should not have year 2000 or Z-count roll-over problems. Users of older GPS receivers may need to check with unit manufacturers on what, if any, remedial action is required, either resetting or replacing memory, and ensuring that software reliant on the receiver incorporates time-validation checks.
Joe Gwinn, of Raytheon Electronic Systems, warns that "-it is not proven that receivers will take the roll-overs in their stride - some receivers may claim wrong locations".
If GPS receivers cannot provide continuous time data, then the systems in which they are installed must be able to validate the data stream. The only way to ensure that a system can do this is by using a JPO GPS simulated signal. For systems where total validation is impractical, but reliability essential, some observers believe that it may be prudent to move towards combined GPS/Glonass receivers. The Glonass has similar problems to those of the GPS, but at different times, thus providing some emergency cover.
A widely distributed memo from the JPO requests "-assistance in ensuring that a phenomenon associated with the year 2000 does not cause catastrophic mission failure in your GPS integration-we are concerned that the 'daisy-chain' implications may not have been fully investigated by platform integrators". The JPO hopes that all JPO-procured receiver designs will be compliant this month, and is making testing available to GPS developers on a "fee/non-interference basis". It is then up to any suppliers of non-JPO-procured equipment to right any problems.
"Embedded" processor systems, such as those used in flight control, ATC and military-type command, control and communications (C3I) systems, represent the most safety-critical software applications, and installations are often in inaccessible locations. US company Mitre puts the repair cost for such devices at up to $8 per line of code checked. Each system has thousands, or even millions, of lines. The highest cost is likely to be incurred where validation involves testing in a real environment with, for example, aircraft being test-flown for their own systems and ground systems. This cost is likely to climb as the clamour for programmers grows.
Importance of civil GPS
On the ATC front, the US Department of Transportation (DoT) now has authority over increasingly important civil GPS use in the USA, as terrestrial navigation systems are being decommissioned (the Omega system will be shut down on 30 September, for example). Airlines, increasingly conscious of the cost penalties of not being able to operate in a free-flight world, are switching accordingly to the GPS, with the attendant pitfall of over-reliance on the system.
The US Federal Aviation Administration aims to conclude its assessment stage for all of its systems by December. It will then move to a "renovation" stage and, finally, to testing, which it estimates at around half the total effort. Because of the limited time remaining, systems are being prioritised, with mission-critical devices at the head of the queue. Some systems have been running without maintenance for years, so the source code may be lost, and will thus need rewriting. All commercial off-the-shelf software is being referred back for modification and independently tested, and must follow strict year 2000 guidelines. There is some suggestion that a contingency plan will need to be in place, however, to provide support if ATC glitches should occur.
As far as airlines are concerned, there are many other facets of their businesses which may be vulnerable to the problem. Companies such as Unisys have introduced programmes for its airline customers to help them prepare mission-critical applications for the year 2000, including computer-reservations and freight-management systems.
Source: Flight International
