SEPTENTRIO PolaRx5TR, Reference Manual

Page: 69 / 512

2 Operation Details
sure to enable the OSNMA operation at the same time (using either the
loose
or
strict
argument).
2.7 Receiver Autonomous Integrity Monitor-
ing (RAIM)
The receiver features RAIM to ensure the integrity of the computed position solution, pro-
vided that sufficient satellites are available. The RAIM algorithm consists of three steps: de-
tection, identification and adaptation, or shortly “D-I-A”
(3)
:
• Detection : an overall model statistical test is performed to assess whether an integrity
problem has occurred;
• Identification : statistical
w
-tests are performed on each individual measurement to
assess whether it should be marked as an outlier;
• Adaptation : measurements marked as an outlier are removed from the position com-
putation to restore the integrity of the position solution. This step is only applied if
outliers have been detected in the detection step.
If the overall model statistical test fails, the RAIM module attempts to recover from the in-
tegrity failure by removing the responsible measurement(s) identified in the second step.
As a consequence, the RAIM module will generally increase the continuity of integrity. The
IntegrityFlag
field of the
RAIMStatistics
SBF block reports an integrity failure if in-
sufficient measurements remain after outlier removal (after several D-I-A steps), or if the
overall model statistical test fails while no outliers can be identified. In the latter case the
"sum of squared residuals too large" error is reported in the
Error
field of the PVT related
SBF blocks.
The statistical tests assume an a-priori model of the measurement error probability distri-
bution. As such, these tests can have the four classical outcomes in hypothesis testing, as
shown in the table below (the letters A, B, C and D refer to the samples in Figure 2-5):
no outlier outlier present
False Alarm Correct
outlier detected
(type I error)
A D
Correct Missed Detection
no outlier detected
(type II error)
B C
The RAIM module makes a correct decision in two cases: an outlier present in the data is
indeed detected, and no outlier is detected when none is present. However, when no outlier
is present and the RAIM module declares an outlier is present, a false alarm is triggered.
When an outlier remains undetected, a missed detection occurs.
(3)
Baarda, W., A Testing Procedure For Use in Geodetic Networks, Netherlands Geodetic Commission, Publ. On
Geodesy, Vol.2, no. 5, 1968
69