Doomsday for GPS receivers?Jan 1, 2003
Many computer users are aware of the danger that the year 2000 poses for computer records, operating systems, and software applications that use a two-digit system for signifying date (i.e., DDMMYY). GPS, it seems, will experience a similar watershed event when the GPS week count does a "rollover" at midnight on August 21, 1999.
The rollover could cause some GPS receivers to think that August 22, 1999, is actually January 6, 1980. While this date error sounds serious, it shouldn't cause problems for most GPS receivers when it comes to fixing position. However, the effects of the rollover depend on how the receiver's software has been put together. The possibility exists that, if a receiver manufacturer hasn't taken the rollover into account, then the unit might have problems. "The basic algorithms for computing the position of a receiver do not rely on the calendar date or UTC," said Richard Langley, professor of geodesy at the University of New Brunswick in Fredericton. "So long as the software doesn't rely on the calendar date in any way for position computation, the position computed by the receiver will be just as accurate after rollover as before."
GPS maintains a week count that helps receivers in indicating GPS time and date. The count is part of the overall GPS navigation message. This message is a 25-frame sequence, with each frame composed of 1,500 bits of data that tell a receiver such things as the necessary satellite clock corrections, ionospheric coefficients, satellite orbital positions, plus an indication of the health of each satellite in the constellation. Each frame takes 30 seconds to transmit and the whole navigation message can be received in 12.5 minutes. Each of the 25 frames are further broken down into five subframes of 300 bits.
The week count is signified in by a 10-bit (2 or 1,024) number that increases by one increment each week. Since 1,024/52 = 19.69, the week count runs out of increments after 19.69 years and starts again at zero. The week count began on January 5, 1980, so the 19.69 years are up on August 21, 1999. The next rollover of the week count will take place in 2019. Unless manufacturers of GPS units took this rollover into account when they wrote the operating software for their receivers, there's a chance that when the week count resets to zero receivers will be fooled into thinking that it is actually January 6, 1980.
The wrong date has the potential to disrupt the way in which a receiver gathers the satellite almanac, which is a list of the orbital positions of each of the 24 GPS satellites in the constellation. Receivers can store this list in memory and then use it later for acquiring and tracking satellites. Should a receiver have a faulty date due to the rollover, it might look at the date of the almanac and think that it was old data and discard it. In such a case, a receiver might then have a difficult time acquiring satellites and tracking them. In other words, it might stop providing positions. Without a full-blown GPS simulator, it is impossible to say how the rollover issue could affect a GPS receiver. "There appears to be sufficient information in the GPS signal to allow accurate navigation even at the moment of rollover," said Joe Gwinn, a principal engineer at Raytheon's E-systems division who works on U.S. Navy shipboard electronics systems. "However, unless the receiver firmware programmers were intent on ensuring that their code would operate both through and after rollovers, the code may well fail. More specifically, if the original acceptance tests did not directly address the issue, it is fairly likely that the firmware will stumble at rollover, and beyond. Even if accuracy of navigation is unaffected, seeing the date off by 20 years is likely to destroy people's confidence in the navigational data provided by GPS and, thus, in GPS itself."
How this will affect you with your particular GPS unit will depend on what brand and model of receiver you have. According to John Howie at Leica, some Leica and Magnavox (Leica bought Magnavox's GPS division in 1995) units will need to be updated. "It will require a software update," said John Howie, manager for marine products. "Users can send their units to us or bring them to a marine dealer for the change."
Other manufacturers may still be assessing their product line to determine which of their units will be affected. Raytheon is reportedly looking at its current and former product line. "We are alert to the fact that we have to check the software and the products in the field," said Jack Trommer, manager of design integration at Raytheon.
One manufacturer, Trimble Navigation, has tested its three models of basic GPS sensors (the sensor is the receiver "engine" around which a GPS set is built). "From the sensor point of view, we're in pretty good shape," said Ralph Eschenbach, chief technology officer at Trimble. "We've tested all of the sensors and they all will handle the rollover." According to Eschenbach, a bigger question is how the rollover will affect the types of products in which the sensors are used. "You get into the question of how it will handle the display and how it will handle communication with the user. There may be a point after the rollover when a receiver doesn't gather the almanac properly. The worst thing that would happen is that you'd have to reset the receiver."
Some GPS receiver makers are saying that all of their products have the needed software to handle the rollover. "We recognized the issue way back when we first started designing receivers in 1989," said Cliff Pemble, software engineering manager at Garmin. "We designed our equipment to handle the rollover." Similarly, Norm Cantin, engineering manager at Northstar, said, "It takes a little more software to deal with the rollover. But it was something in the spec. that we took into account."
So, while you should contact the manufacturer of your receiver to check on how it will handle the rollover, it seems that many GPS units will be able to take in stride the resetting of the week count.