AF447

[JD-EE]

Regarding GPS and vehicle status information

I've noticed a fair number of non-pilots mumbling about GPS as if they think it tells you everything about where the plane is and how fast it's going. It's time to put some details on the record.

First, GPS tells you nothing about where you are. It tells you where you were at the time of the last measurement. With an automobile we're speaking of discrepancies that are not worth thinking about. With a high dynamics aircraft (fighter) the difference between where you are and where GPS alone tells you can be more than slightly significant. GPS alone even has difficulty tracking through typical military fighter aircraft moves. With C/A code tracking it's just about worthless. That is why the military has its P-code (and encrypted Y-code).

This lag and inability to track is mitigated to a high degree via software and other inputs, the IMU and IRU for example. Kalman filters can generate enough accuracy to feed back a position estimate to the GPS receiver that the receiver can maintain lock, if not accuracy.

Still, the whole system tells you where you were and where it thinks you might be based on projections not where you are.

GPS is even worse at telling you your speed, although it can do so. It can take where you were at two times and divide by the time to get a velocity vector. If it smooths that over time it can be reasonably accurate.

That brings up "time" as your enemy. The longer you integrate the GPS tracking feedback loops (the lower their bandwidth) the more accurate the data. Noise is the enemy. Fortunately GPS is designed to work WITH jamming efforts. And jamming is just another word for noise. So there is a potentially very large margin for unjammed GPS to be very accurate, especially in relative positions over a short time frame.

Now consider the plane is in heavy turbulence. The filters are prepared to handle this. Civilian aircraft don't do aerobatics. They fly sanely, even in turbulence the flight is fairly sane, it cannot be called aerobatics with spins. rolls, and other rapid motion changes. So far the GPS plus other information sources can give you reasonably good estimates of where you're going, how fast, and where you are. To be sure it's accuracy is fading. But you are within the design parameters for all the instruments and software involved.

Note that I've not said a thing about heading, yaw, or pitch of the plane itself. GPS CAN make an estimate if you have antennas on the nose, tail, and both wing tips. It might even be a good guess using phase tracking tricks. Theoretically it's possible. And with computers these days it may even be practical if calibration tables for cable measured temperature versus delay time is carried and usable. I'd not bet on it.

Now throw the plane into a spiral dive or a flat spin or almost any other controlled or uncontrolled aerobatic movement. Civilian GPS is not designed to handle this very well. The nav computer can make some estimates. But I question whether the dynamic ranges of the instruments, including GPS, can handle the uncontrolled "aerobatics" situation. I'd also be surprised if the SatCom antenna aiming algorithms can adapt quickly enough to the questionable data to keep the antenna aimed on the satellite.

That's as far as I can take it. This is stuff I know about. It is far enough to suggest strongly, but not conclusively, that the plane was still relatively speaking straight and level at 0214Z, the time of its last transmission. Now, the time gap MIGHT mean it was more or less out of control and the antenna aligned for a few hundred milliseconds, long enough to spit out a burst of trouble reports. But that magic lineup seems to me to be to low a probability to worry OUR minds about. The accident investigators might just to kill even the remote possibilities.

{^_^} Joanne - worked on GPS satellite hardware and software and worked on some GPS receiver software in her years at Rockwell and Magnavox (military) in Torrance.

(It's Kalman not Jalman, you bimbo. {^_-} Typo fixed.)