Final Report of BeJamDetect Project 
‚Sb 
LA 
recorded independently of the last nominal recording time, and then it enters a dead-time period 
af 30s to prevent multiple recordings while the system is under interference. The 1-hour counter 
and the dead-time are implemented for the sake of reducing data size in storage. After the end of 
the dead-time, data recording is be resumed, and the system either records again in case of RFI 
detected or resumes the 1-hour counter in the absence of detectable RFI. 
The RFl-detection and recording software is implemented on C++. A circular buffer for the IF 
samples that is realized in the FPGA-part is controlled by the recorder software running on the CPU 
af the platform. The control software was designed to detect the presence of interference and 
trigger recording of snapshots of real-valued IF samples for further analysis in post-processing. A 
single signal snapshot is defined as a block containing the signal samples of all seven array elements 
an both frequency bands (i.e. L1/E1 and L5/E5a). The sampling rate of the IF samples is 100 Msps 
Which results in the digital IF of 25 MHz. Each signal sample is quantized with 14 bit resolution. 
The length of the snapshots can vary between several milliseconds until hundreds of ms. With the 
Demonstrator 1, the length of 30 ms (96 MB) was used. The power level of the RF front-end outputs 
is adjusted in such a way that only the amplitude range of 3-4 bits is used during the quantization 
orocess under interference-free conditions, while the full amplitude range of 14 bits is utilized to 
‚esolve strong RFI signals without suffering from clipping effects. The recorded signal data are then 
transferred to a network attached storage (NAS). 
Interference detection metrics and analysis methodology in 
post-processing 
The statistical test used to detect interference was the M-of-N test. Its adoption is motivated by the 
fact that: due to the very low-power levels of satellite navigation signals, below background noise 
level, the statistical distribution of samples approximately follows the Gaussian shape of white 
noise. It is also the case because of the spectrum spreading modulation of the navigation data by 
oseudo-random noise (PRN) codes. These are deterministically generated binary sequences that 
have statistical properties very similar to white-gaussian noise [7]. 
On Figure 3-6a, the histogram of the samples from the ADC of the fifth antenna array element in 
nominal case (without interference) is displayed. The bell shape close to the gaussian distribution 
zan be clearly noted from the distribution of the 55 bins. In the histogram, samples of approximate 
digital amplitude - which could be translated as a value proportional to the actual power of that 
sample - are grouped in a bin whose height is equivalent to the number of samples in that bin 
Thus, it can be considered that if too many received navigation samples lie beyond specified limits 
af the gaussian, it might be due to the presence of an overlaying signal of higher-power, le. an 
interference signal. This test metric belongs to a family of interference detection techniques that 
‚nvestigate the probability density function (PDF) of the received signal [8, 9]. These techniques are 
more efficient against jJammers, because they usually transmit high-power signals to accomplish 
averwhelming the navigation signals, in contrast to spoofers that attempt to mimic them. 
Fitle: Final Report 
Version: 1.0 
302: