How Accurate Is GPS?

Even the simplest civilian Global Positioning System (GPS) receivers are capable of ranging precision under 10 meters, which translates into a horizontal-positioning precision of 15 or 20 meters. But these are average values, and we need to remember that precision (the fineness of the ruler's scale) and accuracy (how true each marking is) aren't the same.

GPS accuracy is affected by two external factors. First, physical or environmental influences can alter the paths of GPS signals. The receiver assumes that the radio signal has traveled at the speed of light, and it converts a time measurement to distance (like estimating the distance to a lightning bolt from the speed-of-sound delay on the thunderclap). I f this propagation speed is changed by the atmosphere, an error is introduced. Civilian (i.e., nonmilitary) receivers use only a crude model to account for ionospheric variations.

A second, more severe inaccuracy is caused deliberately by the operators of the GPS, who introduce intentional fluctuations on the timing of the "civilian" signals. Without these fluctuations, euphemistically called selective availability (SA), state-of-the-art civilian receivers could be almost as precise as military receivers, which use different, encrypted signals from the same satellites. The Department of Defense considers this possibility dangerous to national security and imposes SA to limit positioning accuracy to 100 meters or better, 95 percent of the time, averaged worldwide. (See "A Stationary GPS Position" for a graphic depiction of SA's effect.) Part of the variation may be due to signal-propagation variations, but most of it comes from SA.

Both SA and i onospheric errors can largely be compensated for with differential corrections, where a fixed monitoring station makes GPS measurements and transmits them to the user over some form of radio link. Differential corrections basically calibrate each GPS signal path against a known reference location. It's like telling the user that when his or her tape measure reads 900 meters, it's really 898. The user can then infer that the 902-meter mark really means 900.

An interesting sidelight on SA is rumored to have occurred during the 1991 Gulf War, when U.S. military forces had only a limited number of military-grade GPS receivers that could use the more accurate position information. Their solution was to turn off the SA scrambling, so that troops could use readily available civilian-grade receivers.

---

Selective Availability Affects Accuracy

illustration_link (34 Kbytes)

The effects of selective availability on positioning accuracy are shown here. A stationary GPS receiver took 6720 measurements at 1-second intervals. While the unit didn't move, the indicated position certainly did. The receiver was located in North London, U.K., at 51 degrees 36.8496' N, 0 degrees 11.149' W. Note that the majority of positions are within 50 meters of the origin.