IEEE - Aerospace and Electronic Systems - March 2021 - 49

Feng et al.
Table 3.

New GPS NAV message received in channel 6: subframe 5
from satellite GPS PRN 26
Position at 2019-Nov-30 10:59:42.000000 UTC using 4 observations is
Lat = 47.251759020 [deg], Long = 5.993861290 [deg], Height =
687.019 [m]

Spooﬁng Cancellation Capability As A Function of
Spooﬁng Signal Power
Power

Constellation

(dBm)

Correct
pos.

Wrong
pos.

No
solution

(%)

(%)

(%)

none

Current

100/100/
100/96

Not
relevant

0/0/0/4

À35

Current

0/90/100/99

57/0/0/0

43/10/0/1

À40

Current

0/93/100/99

96/0/0/0

4/7/0/1

À45

Current

0/2/100/100

61/1/0/0

39/97/0/0

À50

Current

0/3/100/99

31/7/0/0

69/90/0/1

À55

Current

52/23/100/0

0/0/0/0

48/77/0/
100

À60

Current

88/64/100/
13

0/0/0/0

12/36/0/
87

À40

-6 h

7/100/100/
100

44/0/0/0

49/0/0/0

À50

-6 h

6/4/100/32

90/96/0/0

4/0/0/68

Each entry is displayed as percentage of solution " before
correction " / " after cancellation using the LS method " / " after cancellation using the proposed method " / " after cancellation using
the method described in17. "

single-board computer Raspberry Pi4 as a front-end processing of the signal source of gnss-sdr, and while the
results presented here are obtained by postprocessing, similar data have been collected following real-time processing
using the modified source code available at https://github.
com/oscimp/gnss-sdr. Compiling gnss-sdr to Linuxbased embedded platforms using the Buildroot framework
is described at https://github.com/oscimp/PlutoSDR/tree/
for_next.
Since variable initialization in gnss-sdr is random,
the capacity to recover the authentic constellation information is a statistical result: In all following experiments,
a given dataset is decoded 100 times with the same configuration script tuning gnss-sdr parameters, so that successful identification of the authentic signal is a statistical
information given in percentage. Decoding the navigation
messages needed for positioning requires longer records
than the 10 s authentic or spoofing signals shown previously: All signals collected for demonstrating spoofing
suppression by recovering the authentic position in the following analysis are 3-min-long records.
A successful decoding results in the following messages:
New GPS NAV message received in channel 15: subframe 5
from satellite GPS PRN 04
New GPS NAV message received in channel 13: subframe 5
from satellite GPS PRN 02
New GPS NAV message received in channel 3: subframe 5
from satellite GPS PRN 23

MARCH 2021

Velocity: East: -0.092 [m/s], North: -0.091 [m/s], Up = -0.207 [m/s]

in which positioning subframe navigation messages
have been successfully tracked and the true receiver position decoded. Failing to cancel the spoofing signal results
in an erroneous latitude/longitude field, whereas failure to
decode any information is identified with the lack of Lat/
Long fields after processing the 3-min-long records.
Table 3 summarizes the result of running gnss-sdr on
various datasets in which a pair of antennas is exposed to a
clear-sky view of the genuine constellation and spoofed
over the air by a signal transmitted by the PlutoSDR with
a power indicated in the first column, with " none " referring to the absence of spoofing (reference measurement).
In each processing case, we indicate how many times the
correct position is decoded, the erroneous spoofed position, or no solution is found after processing the 3-minlong record. The four fields in each column refer to the
raw collected data (one of the two antenna dataset processed by gnss-sdr), cleaning the dataset to cancel spoofing using the LS method described in (33), cleaning the
dataset to cancel spoofing using proposed method as
described in (20), or cleaning the dataset to cancel spoofing using the reference published method described in
[17] and summarized in (8) and (25).
By analyzing the results exhibited in Table 3, we can
conclude the following.
(a) Suppressing the spoofing signal by the proposed
method always leads to signal recovery in this set of
experiments, both strong and weak spoofing signals
can be effectively suppressed.
(b) At strong spoofing power (À35 and À40 dBm), the
LS-based method allows for identifying the spoofing signal weight, and hence, the spoofing signal
can be suppressed, corresponding to the results
demonstrated in Figures 9 and 12.
(c) At weaker signal (À45, À50, and À55 dBm),
authentic and spoofing signals compete so that
erroneous positions are detected prior to spoofing
suppression. Although the LS-based method can
still act as the reference for the proposed method,
it is unable to properly recover the authentic signal
and failure to locate the receiver is the most common result.
(d) At low power (À60 dBm), spoofing cannot work well,
the correct position can be reached without spoofing
suppression in some cases, whereas, by using the proposed method, the actual position can be reached with
a 100% percentage.

IEEE A&E SYSTEMS MAGAZINE

49
https://github.com/oscimp/gnss-sdr https://github.com/oscimp/gnss-sdr https://www.github.com/oscimp/PlutoSDR/tree/for_next https://www.github.com/oscimp/PlutoSDR/tree/for_next
IEEE - Aerospace and Electronic Systems - March 2021

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