Final Report of BeJamDetect Project
10*
ADC bin histop-am. L1 band, antenna 5
1 Sample bins
Thresholde
m}
ADC bin
104 ADC bin histogram, L1 band, antenna 5
' Sample bins
Threchal—-
11
ADC hir
0
OU
„A
DLR
a) Nominal case from June 26, 2021 at 10:21:43 MET b) RFI case from June 26, 2021 at 11:30:03 MET
Figure 3-6: ADC bin Histogram
In contrast to Figure 3-6a, on Figure 3-6b the histogram of the samples coming from the fifth array
alement clearly depict the usage of bins deviating from the usual Gaussian shape. This means that
more high-power samples were received than in the nominal case. The large lateral lobes (Figure
3-6b) indicate the presence of continuous-wave interference.
The limits were tuned for each antenna based on measurements performed on the system in DLR
in order to cater for the specific characteristics of the hardware. The tuning was also performed
considering the number of outliers that would trigger an alert of interference presence, le. the
1umber of samples beyond the inferior and superior limits of the Gaussian shape.
According to this statistical test, interference is flagged if the number of samples in a block of N
samples outlying the stablished limits of the gaussian surpasses a threshold of M samples. In this
project, M is equal to 400 and N is equal to 1500. A moving window sweeps through the entire
length of the data with an overlapping factor that cause some part of the data in the block
generated by the previous window to be also accounted in the following block. In the project, the
factor was equal to half of N, i.e., half of the data in one block is accounted in the next block.
Among all the data blocks of one snapshot, if at least one block presents more than 400 outliers,
an interference flag is triggered. Interference is finally considered detected if the M-of-N metric
accuses interference in at least two antennas, else it is considered nominal.
To illustrate, on Figure 3-7 are presented a nominal case, in which all antennas have data below
the threshold (400 samples), and an RFI detected case, in which more than 400 samples of the
ADCs of antennas 2 and 1, each, are outliers in the Gaussian shape.
Fitle: Final Report
Version: 1.0
302!