Aerospace and Electronic Systems Magazine November 2017 - 48

ADS-B Jamming Mitigation
first source, in the temporal windows
in which there is not any overlapping.
This is done by exploiting the SVD.
After mixing the matrix estimation,
the whole signal space can be divided
in two subspaces: the first one of dimension one corresponding to the first
source and the other one (orthogonal
to the first one) of dimension m − 1
containing the residual sources, i.e. the
second signal. Estimation of the mixing
matrix doesn't need any array calibration (as in the case of classical beamforming) and can be recursively implemented to divide the original space up
to m subspaces (where m is the number
of antenna elements/receivers).
More in details, the EPA algorithm's steps are the following (see also
Figure 3):

Figure 3.

Example of overlapped sources and EPA Steps.

JAMMING MITIGATION TRIALS

1. A sliding time window (containing 400 samples, i.e. 8 μs long,
being the sampling frequency equal to 50 MHz) is applied to
data X; it is continuously shifted by 2 μs steps;
2. The SVD is applied to the data in the window to obtain the
estimated number of sources versus time. This is achieved by
counting the number of singular values greater than a noise
threshold computed knowing the noise distribution;
3. If there is at least one time slice Xi with only one source (i.e.
only one singular value greater than the threshold), the corresponding singular vector is used to extract (by projection) that
source. Denoting the one-source time slice as X1, the pertaining
steps are the following:
i. Let U be the left-singular vectors matrix obtained by the
SVD of X1, UΣV * = SVD( X1 ), and let u1 be the first row of
U. The subspace belonging to X1 is estimated by the projection X · u1, and the related source is fed to the ADS-B
decoder;
ii. The residual subspace, containing the other sources, is obtained in the same way using Ud−1, defined as U = u1 , U d −1  T
4. If the dimension of the residual subspace Ud−1 is unitary, the
related source is fed to the decoder, otherwise the residual subspace is fed into step 1.
After this processing, the receiver can decode up to m signals
containing up to m independent source (it is assumed that the
sources are spaced and have different azimuth angles). The jamming signal will be not correctly decoded and the legitimate signal
will be not corrupted from the jammer any more, and correctly
decoded.
More details about the algorithm, its derivation, and its performance in term of garbling mitigation and channel throughput
can be found in [28]. In the following, we will concentrate on the
performance of this algorithm to mitigate the jamming on an ADSB message.
48

To verify the performance of the algorithm, we used the four linear
channels of the TDR receiver, injecting different types of digitally
generated jamming signals at the IF stage after the sampling of the
signals as shown in Figure 2.
This method is chosen for the following reasons: (a) to easily
set the desired SIR for each ADS-B message, (b) to set the AOA
of the jamming signal (in this case randomly generated from 0 to
2π) without adding more hardware to the test bed (e.g. splitters and
phase shifters for jammer signal distribution to the four receiver
channels), and (c) to avoid the radiation of jamming signals in the
air.
The unique drawback of this solution is that the jammer signal doesn't experience the real channel propagation losses (e.g.,
temporal dispersion, spatial dispersion, and spectral dispersion) as
the ADS-B signal coming from a real airplane does. In the first approximation, this is acceptable due to the fact that usually the jammer is placed near the receiver with a velocity that is zero or near
to zero. A channel propagation model, such as the model proposed
in [29], can be introduced in the future for the injected jamming
signal to improve the faithfulness to reality of the trials. On the
other hand, the only other possible solution is to radiate the jamming signal.

Spoofing: The intentional
transmission of signals in
the same band as the ADS-B,
with the same protocol but
with erroneous information
displaying false aircraft.

IEEE A&E SYSTEMS MAGAZINE

NOVEMBER 2017

Table of Contents for the Digital Edition of Aerospace and Electronic Systems Magazine November 2017