213 
The DGPS corrections for the test have been obtained in 
three different ways: 
a) by the OMNISTAR system, in real-time mode; 
b) using as base receiver the GPS permanent station of 
Perugia University (mean distance from rover about 100 
km), with a post-processing solution; 
c) assuming as base point another GPS station closer to 
the rover (in Ancona: distance from rover 0 to 25 km), in 
post-processing. 
First of all, a phase double differences kinematic solution 
has been computed to get a reference result. On that 
purpose, the well-known Geotracer OTF software by 
TerraSat has been used. The mean accuracy of this 
reference solution can be estimated as a few centimetres 
over the entire path. 
Thereafter, three different DGPS solutions have been 
calculated in the three ways referred above. A compa 
rison between the phase solution and the DGPS results 
has been finally carried out, analysing the differences 
found. 
The differences between the three different DGPS 
techniques used are rather small, confirming the 
substantial equivalence of the OMNISTAR "virtual base 
station" to a real base station located at some tenths of 
kilometres from the rover. 
In figure 8 are shown the RMS of the rover co-ordinates in 
the OMNISTAR solution, in the path Ancona-Jesi. Below 
that, figure 9 shows a plot of the differences between the 
OMNISTAR DGPS solution and the OTF one. Assuming 
the OTF solution to be "true", one can notice the DGPS 
path is rather "noisy". For most epochs, however, the 
coincidence level between DGPS and OTF is quite good, 
with differences less than 2 meters. Still, there are many 
error peaks, some of which give more than 10 meter 
difference in position. 
The two major gaps in the plots correspond to road 
tunnels situated along the path. A relatively "quiet" period 
is that between the epochs 476300 and 476700, 
corresponding to an almost horizontal road without 
obstructions. The first part of the path interested an urban 
zone of Ancona town. The last part, after the second 
tunnel, is a hill road, with frequent curves and trees on the 
road sides. 
From the above results, the necessity of an appropriate 
validation or filtering procedure for such DGPS kinematic 
outputs is more than evident. 
The b) and c) solutions have also been obtained using a 
phase-smoothed pseudorange procedure. This way the 
results have been much less noisy than in DGPS mode, 
with a coincidence with the OTF solution at about 10 cm 
level. 
general considerations can be formulated: 
• Good results (accuracy about 1 meter or slightly less) 
can generally be achieved through the OMNISTAR 
DGPS technique under good field conditions (visibility 
of 6-7 GPS satellites; continuous reception of RTCM 
signal); 
• Field operations are very simple, but a screening and a 
validation of the DGPS output is advisable, particularly 
for kinematic applications; 
• In static mode, a relatively easy validation procedure 
can be based on averaging the results obtained during 
the permanence on a given point; this way, the 
position accuracy can be improved to some tenth of 
centimetres; 
• In kinematic mode, specially when frequent 
obstructions are present along the path, appropriate 
filtering and validation procedures have to be studied 
and experimented; an application of the Kalman 
filtering could be suggested. 
REFERENCES 
Dominici D., M.L. Pecetti, F. Radicioni, A.Stoppini (1999), 
Analysis and validation of different DGPS techniques. 
Proc. of DGPS Trieste Meeting, Trieste, March 1999. In: 
Reports on Geodesy, Warsaw University of Technology. 
Note: 
The researches reported on this paper have been partially 
financed with MURST funds pertinent to COFIN97 project. 
4. Final remarks 
On the basis of the experiences carried out until now at 
Perugia University on the OMNISTAR DGPS, some