210 
ground, off centre with respect to the antenna installation. 
Contemporarily to the GPS session, the antennas have 
been surveyed from the off-centre marker, using a 
reflectorless laser EDM with an electronic compass and an 
electronic inclinometer, mounted on the same vertical as 
the GPS antenna (fig. 5). 
Fig. 5 - Equipment for the static test: 
1) GPS receiver connected with external OMNISTAR kit; 
2) GPS antenna; 3) OMNISTAR antenna; 
4) reflectorless EDM; 
The GPS measurements have been carried out with a 
double frequency Novatel receiver, also using the phase 
data registered by the Perugia University GPS permanent 
station. The maximum distance between this reference 
stations and the surveyed points was about 40 kilometres. 
Fixed double differences baselines have been computed 
from the Perugia station to all surveyed points, in order to 
have a good reference solution to which compare the 
DGPS results. 
Sessions of about forty minutes have been effected, in 
order to have a sufficient number of DGPS solutions for a 
good statistical analysis. Sampling interval was 10 
seconds. 
Table 1 (printed on the following page) shows a series of 
DGPS results registered at the beginning of a typical static 
session. The series starts with the turning on of the 
receiver. We can notice the first 8 epochs are 
characterised by a rather high RMS (some tenths of 
meters) for all three components East, North and Height. 
Since the 9th epoch, the RMS instantaneously decreases 
to about 1 meter. 
The reason is evident: no DGPS correction for the first 8 
epochs, then the RTCM signal has been received, 
decoded and used by the position computing software. 
This has required about 1 minute and 40 seconds from 
turn-on. Only the first 30 epochs of the session are 
contained by the table; the rest of the session has given 
quite similar results, without error peaks. 
The positions obtained in DGPS mode have been 
compared to the results of the fixed double differences 
solution, the latter assumed as a reference. The 
comparison is synthesised by the plot in figure 6, referring 
to the same values of table 1. 
The uncorrected solutions corresponding to the first eight 
epochs are clearly recognisable on the east side of the 
plot, with an offset of about 25 meters from the double 
differences position. The further epochs' positions form a 
"cloud" of points around the static GPS solution, with a 
maximum diameter of about 2.2 meters in north-south 
direction. The RMS is about 50 cm for the North 
component, 30 cm for the East. The mean of all DGPS 
positions differs from the static solution of about 20 
centimetres. 
Quite similar results have been obtained for all other 
points included in the test. 
From the test results, we can conclude that: 
• the order of DGPS accuracy has been about 1 meter, 
substantially confirming the company specifications; 
• a better solution (20-30 cm accuracy) has been 
obtained by simply averaging the instantaneous results 
registered during a static DGPS session of some 
tenths of minutes. 
It is to be remarked that our test was performed with a 
good geodetic 12-channel GPS receiver. Using simpler 
receivers, down to code-only type, the DGPS accuracy is 
expected to get worse (a few meters RMS) because of the 
higher receiver noise. 
3.3. Tests in kinematic mode 
Other tests have been performed in kinematic mode. The 
example here presented refers to a road path from 
Ancona to Jesi (central Italy) and back, travelled by car 
(fig. 7). The path length was about 25 kilometres. The 
GPS mobile equipment is substantially the same of the 
former test: a double frequency Novatel receiver 
connected to an OMNISTAR external kit with its own 
antenna.