IEEE - Aerospace and Electronic Systems - April 2021 - 41

Dautermann et al.

Figure 4.
Ground track recorded during the flight trials on February 14,
2020.

collect data for postprocessing only and we were not able
to transmit a live GLS signal from the ground. We
recorded GNSS data including SBAS using a Septentrio
PolaRx3 receiver at 10 Hz from an experimental GNSS
L1/L2/L5 multiband antenna installed on top of the aircrafts fuselage.
At 09:56 UTC, the aircraft began to conduct three RNP Z
approaches to Runway 34 at Thessaloniki. The ground track,
waypoints and terrain are shown in Figure 4. The approach
began from the west at the initial approach fix waypoint
APZOC with a minimum altitude of 5000-ft MSL. Next, the
aircraft passes the intermediate fix CEFEB at 4000-ft MSL
or above. The final approach commences at 3500-ft MSL at
the waypoint TS626, from which the aircraft descents on a
3.8 path toward the runway for landing. The final approach
is rather steep and follows the terrain contour of a hill located
south of Thessaloniki airport. During the first two
approaches, the pilots initiated a missed approach at the decision altitude followed by radar vectoring to APZOC as provided by the air traffic controllers. The last approach was
concluded with a successful landing on runway 34.
Since we could not use the installed Collins GLU-925
multimode receiver (MMR to receive a live signal in space)
due to the absence of the frequency permission, we postprocessed the Septentrio data with corrections generated by the
GLASS system algorithms and a multimode receiver software. Figure 5 shows the result of this postprocessing. The
top panel of Figure 5 shows the vertical integrity data
APRIL 2021

calculated by the MMR in GBAS mode as described in [4]
and [12]. As the vertical component is always the more
restrictive one and the horizontal and lateral does not give
additional information, we chose to show here only the data
pertaining to the vertical component.
The top panel of Figure 5 depicts the different Vertical
Protection Levels (VPL) and the alert limit during the flight
test. The green line shows the vertical alert limit. It scales
from 58.75 m down to the final approach segment vertical
alert limit (FASVAL) of 25.4 m. We can already notice that
this value is only achieved at the very end of the trial for a
very brief period of time when the aircraft is below a height
of 60.96 m (or 200-ft) above ground for the final landing.
During the two previous approaches, the alert limit scales
down to 39.4 m during the go-around. This is in accordance
with our argumentation in the previous section, where we conclude that the VAL at the decision height is the limiting factor
and not the FASVAL.
Protection levels (PLs) estimate the position uncertainty
at the allocated integrity risk bound at any measurement
epoch. This calculation is a requirement for any airborne
GNSS receiver (see [3] and [11]). The red line is the standard
SBAS VPL as computed by a pure SBAS receiver certified
according to [11]. The yellow line shows the VPL computed
by the airborne receiver using the GLASS ground system
without any inflation and the purple line is the VPL computed by GLASS ground system including inflation. The
sawtooth patter in the protection levels is typical for SBAS
and takes into account the degradation of the correction
validity over time.
In this snapshot, we can see that the protection levels
behave as expected from Dautermann et al. [3]. The VPL
obtained by the uninflated GLASS system is slightly larger
than the pure SBAS VPL due to addition of the along
track error component to the vertical caused by the aircraft
descending on an angled glide path and the larger K multiplier required by GBAS. Finally, of course, the inflated
GLASS is the largest. When the aircraft is in a turn (at
10:05, 10:17, and 10:29), tracking to some low elevation
GPS satellites is lost and the protection levels increase.
The middle panel shows both localizer (blue) and
glide slope (red) deviations in degrees calculated from the
FAS data block in angular units. The deviation calculation
is stopped at full scale deviation, which is determined by
taking 0.25 times the glide path angle for glide slope deviation and the course width at threshold from the FAS data
block. Arrows indicate the respective value in the panel.
Finally, on the bottom panel, we can see the aircraft
altitude above the WGS84 ellipsoid. Whenever the aircraft
was in the precision approach region defined by [12], we
shaded the plots with gray background color.
Figure 6 shows a magnified view of the last approach. In
the top panel, the zig-zag pattern of the protection levels is
now clearly visible. Contrary to Figure 5, we show the localizer and glide slope deviations in Figure 6, bottom panel, as

IEEE A&E SYSTEMS MAGAZINE

IEEE - Aerospace and Electronic Systems - April 2021

Table of Contents for the Digital Edition of IEEE - Aerospace and Electronic Systems - April 2021