IEEE - Aerospace and Electronic Systems - March 2021 - 46

Software-Deﬁned Radio Implemented GPS Spooﬁng and Its Computationally Efﬁcient Detection and Suppression

Figure 6.

Figure 7.

Minimum phase difference between two antennas of the detected
frequency peaks without (top) and with (bottom) backward sliding average process. Authentic and spoofing signals are alternately sampled with a duration of about 180 s. The red horizontal
lines denote the spoofing detection threshold, different for top and
bottom subfigures, below which the spoofing is determined. On
the contrary, above which there is no spoofing.

Detection results with (top) and without (bottom) backward sliding average process for moving spoofer case. Spoofing is controlled to be started from the beginning and to be stopped at about
9 min. The authentic satellite signals are sampled from 9 to 12
min. The false alarms caused by the noise and the unexpected
large phase differences among spoofing satellites can be avoided
when sliding average process is applied.

differences are the same, a signature characteristic for
spoofing detection.
As described earlier, the power emitted from the static
spoofing SDR ranged from À60 dBm to À35 dBm with a
5-dB step, alternating from spoofing to genuine constellation every 10 s. Figure 6 shows the minimum value of the
successive difference vector DD' (top) computed by (17)
and the minimum value of the averaged successive difference vector DD' (bottom) computed by (19) with an interval of T ¼ 100 ms. In order to calculate DD' , the window
size W is set to 50, i.e., the averaging time is 5 s. It can be
seen from both subfigures that, when spoofing occurs, the
minimum phase difference is much lower than the case
without spoofing. Therefore, the spoofing can be effectively detected given an experiential detection threshold
of 0.04 rad, as denoted by the red lines in Figure 6.
Then, a second experiment was conducted, where the
spoofer emitting À40 dBm is moving on the campus along
a straight path and emitting at constant power for a duration of 9 min and then is stopped to leave the receiver to
sample the authentic satellite signals during 3 min. The
spoofer transmitted at distances between 10 m and 25 m
from the receiving array, resulting in a mix of open sky
and spoofed signal. Similar to the results shown in the
static spoofer case, Figure 7 shows the minimum value
(top) and the averaged minimum value (bottom) of the
successive difference vector with T ¼ 100 ms and W ¼
50. It can be seen that the phase difference is continuously

small in the beginning (from 0 to 540 s or 9 min) and then
becomes big (from 540 to 720 s or 9 to 12 min). Therefore,
the spoofing phenomenon can also be effectively detected
when there is relative movement between the spoofer and
the receiver with an experiential detection threshold of
0.04 rad. After performing backward sliding average process, the false alarms as observed from the top subfigure
(at about 220 s and 520 s) can be avoided, as shown in the
bottom subfigure. Since backward sliding approach is
used, the causal average process is conducted only based
on current and previous phase-difference measurements,
without next measurements, giving real-time spoofing
detection capacity.
As mentioned in " GPS Spoofing Detection " section,
the spoofing detection threshold can be set at a fixed value
in advance empirically as used for plotting Figures 6 and 7,
or adaptively determined by some techniques. For example, for each detection point, referred to as cells, based on
reference cells, a CFAR technique can help to detect the
spoofing phenomenon with a CFAR by dynamically setting
a threshold on the parameter representative of spoofing, in
our case DD'. Therefore, by using the Smallest of CFAR
(SO-CFAR) [28] with a false-alarm rate of 10À6 , a guard
cell number of 50, and a reference cell number of 50, the
detection result for moving spoofing case is shown in
Figure 8. In order to demonstrate the dynamic threshold
selection, the dataset used to generate Figure 7 has been
time-reversed. According to the principle of CFAR,

IEEE A&E SYSTEMS MAGAZINE

MARCH 2021

IEEE - Aerospace and Electronic Systems - March 2021

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