Differential GPS - Understanding GPS: Principles and Applications - page 391

Both receiver noise and multipath errors can change very rapidly. Since these
errors are not common between the user and reference station in a DGPS scenario,
the rates of change of these errors are only important in that averaging of some form
within the user equipment can often be employed to reduce their consequence.

8.3 Code-Based Techniques

Many code-based DGPS techniques have been proposed to provide improvements
in performance over stand-alone GPS. These techniques vary in sophistication and
complexity from a single reference station that calculates the errors at its position
for use with nearby GPS receivers to worldwide networks that provide data for esti-
mating errors from detailed error models at any position near the Earth's surface.
As discussed in Section 8.1, they may be sorted into three categories, local-area,
regional area, and wide area, depending on the geographic area that they are
intended to serve. This section discusses code-based techniques for each of these
categories.

8.3.1 Local-Area DGPS
A local-area DGPS (LADGPS) system improves on the accuracy of stand-alone GPS
by estimating errors corrupting the stand-alone GPS position solution and transmit-
ting these estimates to nearby users.

8.3.1.1 Position Domain Corrections
Conceptually, the simplest way to implement LADGPS is to place a single GPS ref-
erence receiver at a surveyed location, compute the coordinate differences (in lati-
tude, longitude, and geodetic height) between that surveyed position and the
position estimate derived from GPS measurements, and transmit these latitude, lon-
gitude, and height differences to nearby users. For the most part, the coordinate dif-
ferences represent the common errors in the reference and user receiver GPS
position solutions at the measurement time. The user receivers can use these coordi-
nate differences to correct their own GPS position solutions.
Although extremely simple, this technique has a number of significant deficien-
cies. First, it requires that all receivers make pseudorange measurements to the same
set of satellites to ensure that common errors are experienced. Therefore, the user
receivers must coordinate their choice of satellites with the reference station; or the
reference station may determine and transmit position corrections for all combina-
tions of visible satellites. When eight or more satellites are visible, the number of
combinations becomes impractically large (80 or more combinations of four satel-
lites). A second problem may also arise if the user and reference station receivers
employ different position solution techniques. Unless both receivers employ the
same technique, (e.g., least-squares, WLS, or Kalman filters), with equivalent
smoothing time constants, filter tunings, and so forth, position domain corrections
may yield erratic results. For these reasons, position domain corrections are seldom,
if ever, employed in operational DGPS systems.