For proper operation of stage ‘‘Adaptive Kalman Filtering,’’ an
accurate estimation of measurement errors in GPS data is needed.
As such errors are larger, data filtering must be stronger and vice
versa —if the error is small, filtering should be minimal. Errors in
GPS data are of two types (Ogle et al., 2002): (1) systematic errors
and (2) random errors. Systematic errors are mainly due to a low number of visible satellites, a high value of the Position Delution
of Precision (PDOP), and a GPS receiver antenna placement.
Systematic errors are identified and removed easily. The random
errors are the result of shift in satellite orbits, satellite clocks
errors, effects in the troposphere and ionosphere and multi-path
signal reflection. Due to their nature, random errors are difficult
to be identified and filtered. Table 2 shows the average deviation
in the position for different types of random errors (Parkinson and
Spilker, 1996).