FACTA UNIVERSITATIS 

Series: Electronics and Energetics Vol. 33, N
o
 3, September 2020, pp. 459-476 

https://doi.org/10.2298/FUEE2003459M 

© 2020 by University of Niš, Serbia | Creative Commons License: CC BY-NC-ND 

RELIABILITY ANALYSIS OF DIFFERENT RCIED ACTIVATION 

SIGNAL RESPONSIVE JAMMING TECHNIQUES  

AND THEIR COMPARISON TO ACTIVE JAMMING
*

 

Mladen Mileusnić, Predrag Petrović, Vladimir Kosjer, 

Aleksandar Lebl, Branislav Pavić  

IRITEL a.d., Belgrade, Serbia 

Abstract. In this paper we compared the time required for the successful jamming of 

remote controlled improvised explosive devices activation using active and responsive 

jamming methods. As a representative of active jamming method we analyzed jamming 

signal generation using frequency sweep. For the analysis of the possible activating 

signal presence based on responsive jamming procedures we first supposed Fast 

Fourier Transform (FFT) implementation and compared its analysis rate to the rate of 

sweep jamming. Taking into account the current technology state, it is proved that the 

time required to achieve the successful jamming relied on FFT analysis may be less 

than in the case of active sweep jamming. After that we considered pros and cons for 

energy detector and matched filter detector implementation in responsive jamming. For 

these two detector types it is shown how to determine the number of analysis blocks to 

achieve approximately the same number of collected samples as in the case of FFT 

implementation, starting from the probabilities of false detection and miss of detection.  

Key words: Active and Responsive jamming, RCIED - remote controlled improvised 

explosive devices, Frequency sweep, Fast Fourier Transform, Energy 

Detector, Matched Filter, Jamming Reliability 

1. INTRODUCTION 

The common characteristic of all remote controlled improvised explosive devices 

(RCIED) is that they are activated by wirelessly transmitted messages. The results of RCIED 

activation message could be disastrous regarding people lives (VIP persons) and the 

equipments damages. All elements related to activation signal characteristics (signal power, 

frequency, implemented modulation method, message duration) are completely unknown. 

This fact produces great problems in the realization of RCIED activation jammers. 

                                                           

 Received February 4, 2020; received in revised form March 3, 2020 

Corresponding author: Aleksandar Lebl 

IRITEL a.d., 11080 Belgrade, Batajnički put 23, Serbia  

E-mail: lebl@iritel.com 
*

 The earlier version of this paper is awarded as the best one in the section Telecommunications at the 6
th
 IcETRAN 

Conference, Silver Lake, 3-6 June 2019, [1].  



460 M. MILEUSNIĆ, P. PETROVIĆ, V. KOSJER, A. LEBL, B. PAVIĆ  

Contributions [2] and [3] provide a general overview of jammers types, communications 

jamming requirements and their efficiency analysis. Modern communications jamming 

principles and techniques may be found in [4].    

There are two basic approaches to the jammer implementation. The first one is active 

jamming, which consisted of continuous predefined jamming signals sending independently 

of the RCIED activating message characteristics. In this concept there are no „look through“ 

phases to detect the activation message existence and the jamming signal characteristics are 

selected in general using previous experience and expectations. The most important freely 

selected jamming signal parameter is the RF signal level. This level has to be as high as 

possible to successfully prevent activating message reception. Two key features which are 

not optimally chosen relate to continuous jamming regardless of RCIED activation message 

existence and the RF jamming signal level necessary for jamming successfulness due to the 

fact that the activation signal level is unknown. 

The alternative approach to jammer implementation is responsive jamming concept. 

In this case the jamming signal characteristics can be optimized using look through 

intervals to detect the activation message existence and its level. That’s why it is possible 

to send the jamming signal only during activation message presence and jamming signal 

level can be adjusted to the activation message level in order to successfully deny the 

threat. A wide range of active and responsive jammers may be found in [5]-[14]. 

It may be concluded from this short presentation of active and responsive jamming 

characteristics that active jamming is always successful, while responsive jamming 

efficiency depends on activation message detection reliability. The question is whether 

responsive jamming reliability may be higher than for active jamming. In this paper we 

compare the reliability of mostly implemented active jamming method – frequency 

sweep [15]-[19] to the reliability of a representative method for activating signal eventual 

presence detection in order to generate jamming signal according to the activation signal 

characteristics by implementation of Fast Fourier Transform (FFT) in the analysis [14].             

A brief principle schematic of RCIED activation signal detection is explained in Section 

2. After that the method for RCIED activation signal frequency spectrum estimation based 

on FFT analysis is presented in Section 3 with the emphasis on the required time for 

calculation. Sections 4 and 5 deal with the specificities of energy detector and matched filter 

detector implementation for RCIED activation message detection. The emphasis is on the 

determination of collected samples number. Section 6 is devoted to frequency sweep 

jamming and to determination of required time to realize one complete jamming cycle. In 

Section 7 jamming reliability on the basis of FFT analysis is compared to the frequency 

sweep jamming reliability, whereby two special purpose processors are considered for FFT 

calculation. Reliability estimation is based on the required time to allow successful 

jamming. Section 8 is focused on the presentation how to determine the necessary number 

of analysis blocks in energy detection or matched filter detection to achieve the comparable 

sample collection rate of these two detectors to FFT based detector. At the end, the paper 

conclusion is given in Section 9. 

2. PRINCIPLES OF DETECTION PROCESS 

Main principles of RCIED activation signal detection may be explained using 

simplified block-schema presented in Fig. 1. 



  Reliability Analysis of Different RCIED Activation Signal Responsive Jamming Techniques... 461 

 

The first phase in detector function is signal samples collecting (block SCOL). After that 

follows processing of these samples (block PROC). The final step is making a decision 

about (eventual) presence of RCIED activation signal on the base of a set of comparison 

rules (block DECISION). These comparison rules are adjusted to the applied method of 

signal samples processing. 

This paper is mainly devoted to the block PROC. The analyzed methods are FFT, 

energy detector and matched filter. When the second or the third of these three methods 

is implemented, digital filter precedes the phase of processing. 

According to the available literature, there are also other methods which are less often 

applied for spectrum sensing, but they are possible candidates for RCIED activation signal 

detection. Some of them are waveform based detection, eigen-value based detection, 

wavelet based edge detection, ciclostationary feature detection [20] and so on. These 

methods, as generally less often applied ones, are beyond the scope of this paper.      

SCOL PROC DECISION

 

Fig. 1 Block schema of RCIED activation signal detector 

3. SIGNAL SPECTRUM ESTIMATION ON THE FFT BASE 

FFT is the calculation procedure, which allows relatively fast estimation of discretized 

signal frequency spectrum. Starting from n time samples of analyzed signal, this procedure 

gives a snapshot of signal frequency spectrum also in n points, i.e. n spectrum lines are 

obtained. FFT is the optimum method taking into account the required number of 

mathematical operations for signal spectrum determination. There are (n/2)·log2 (n) 

complex multiplications and n·log2 (n) complex additions [21]. The limitation for n is that 

the condition n=2
a
 must be satisfied, where a is the positive integer number. This is a 

significant saving in the number of mathematical operations and in the required calculation 

time comparing to the classical method of frequency spectrum estimation by Discrete 

Fourier Transform (DFT). Namely, it is necessary to perform n
2
 complex multiplications 

and n
2
–n complex additions to obtain n frequency spectrum components by DFT on the 

base of n time samples. 

Let us suppose that fs is the frequency of analyzed signal sampling. The sample 

acquisition time is then: 

 
s

n
t

f
  (1) 

The frequency resolution on the base of sample acquisition time may be determined as: 

1 sf
df

T n
       (2) 

Therefore, frequency resolution is improved when acquisition time is increased, i.e. the 

space between frequency spectral components of the analyzed signal is lower. 



462 M. MILEUSNIĆ, P. PETROVIĆ, V. KOSJER, A. LEBL, B. PAVIĆ  

Constant advancements in processor realization technology and mathematical algorithm 

improvements are visible in two aspects of FFT calculation progress. On the one side the 

number of points in which frequency spectrum is determined is constantly increased, and on 

the other side the required FFT calculation time for some exactly determined number of 

frequency spectrum components is constantly decreased, chronologically, successively 

according to presentations in [22], [23], [21], [24]. We selected two approaches referenced 

in [21] and [24] due to very fast processing algorithms.       

Data presented in [24] is related to the FFT calculation time as a function of the 

number of signal time samples implemented for FFT calculation, i.e. as a function of the 

obtained frequency components number in the analyzed spectrum. The presented data is 

for processor clock of 1GHz. It is further emphasized in [24] that improvement may be 

achieved by processor clock speed increase to 1.25GHz. Besides, it is stated in [25] that 

maximum processor clock frequency may be even 1.4GHz. On the base of these data, the 

FFT calculation time (Tcal in ms) is presented in Table 1 as a function of the number of 

points used in a calculation, for a processor clock of 1.25GHz and for 8 processor cores. 

The value of the constant K is 1024 in the first column of the Table 1. 

The time of FFT calculation (Tcal in μs) according to the data emphasized in [21] is 

presented in the Table 2. The processor clock in this case may be in the range between 

60MHz and 150MHz [26]. That’s why data are presented for the mean processor 

frequency of 100MHz. FFT hardware accelerator (HWAFFT) is one of the parts in the 

processor implemented according to [21]. HWAFFT is intended for faster FFT 

calculation. Data in Table 2 are related to the case when HWAFFT is implemented. The 

number of points is relatively small (till 1024) where FFT is calculated comparing to the 

number of points, where FFT results are presented in Table 1. 

In accordance to Fig. 1, the total time, which is needed for signal analysis in a jammer 

(Tan) before (eventually) starting RCIED activation jamming signal emission, consists of 

three components: sample acquisition time (T), FFT calculation time (Tcal) and the time, 

which is necessary to compare obtained signal frequency components after FFT 

calculation (Tcomp) in order to determine whether it is necessary to start jamming. When 

considering the last component (Tcomp), there is not such a data in a literature, because 

calculation is very specific. For our analysis, we supposed that taking equal values of Tcal 

and Tcomp is a quite good approximation, i.e. 

2an cal comp cal
s

n
T T T T T

f
         (3) 

Table 1 The time of FFT calculation as a function of the number of calculation points for 

the processor presented in [24]  

Number of points for FFT calculation Calculation time Tcal [ms] (8 cores, 1.25GHz) 

16K 0.1051 

32K 0.1584 

64K 0.2517 

128K 0.5128 

256K 0.9488 

512K 2.4824 

1024K 5.1226 



  Reliability Analysis of Different RCIED Activation Signal Responsive Jamming Techniques... 463 

 

Table 2 The time of FFT calculation as a function of the number of calculation points for 

the processor presented in [21] 

Number of points for FFT calculation Calculation time Tcal [μs] (with HWAFFT, 100 MHz) 

8 1.3 

16 1.7 

32 3.21 

64 4.36 

128 9.12 

256 16.68 

512 37.4 

1024 73.15 

4. RCIED ACTIVATION SIGNAL DETECTION BY ENERGY DETECTOR 

Energy detector is the simplest techniques for signal detection [27]. In the same time 

it is a very often applied technique. It is necessary first to measure signal energy in the 

pre-defined frequency band. The measured signal energy is then compared to the energy 

threshold according to the equation 

2

1

( ) ( ( ))

N

n

E x x n 


       (4) 

where N is the number of samples implemented for signal energy estimation, x(n) is the 

amplitude of n
th

 sample and γ is the threshold power.   

Although simple for implementation, energy detector performances are degraded due to 

noise uncertainty (noise level is variable during time) and background interference [28]. 

Noise uncertainty may be bounded or unbounded [29]. As a consequence of noise 

uncertainty, the detection by energy detector may become even impossible under relatively 

low value of signal to noise ratio (SNR) [30]. In other words, there exists a SNR wall: 

detection is possible only when signal power is higher than noise power uncertainty.  

For the analysis in this paper and for the comparison of energy detector characteristics 

with the characteristics of other methods for reactive jamming the most important parameter 

is the number of samples (N) to achieve necessary detection reliability. Our analysis is based 

on the formula for N from [27] [31]:       

1 1 2

2

( ( ) 1 2 ( ))
2

f dQ P SNR Q P
N

SNR

 
   

     (5) 

where Pf is probability of false detection (detector announces signal presence although 

there is no signal), Pd is probability of successful detection and Q
-1

 is inverse Gaussian-Q 

function. In other words, Q
-1

 is the inverse of 

 

2

1
21

( )
2

u

x

Q x e du




  
   (6) 



464 M. MILEUSNIĆ, P. PETROVIĆ, V. KOSJER, A. LEBL, B. PAVIĆ  

1,E+01

1,E+02

1,E+03

1,E+04

1,E+05

1,E+06

1,E+07

-20 -19 -18 -17 -16 -15 -14 -13 -12 -11 -10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0

SNR(dB)

N

Pf=Pmd=0.1 Pf=Pmd=0.01 Pf=Pmd=0.001 Pf=Pmd=0.000001
 

Fig. 2 The necessary number of samples (N) when energy detector is applied as a function 

of signal-to-noise (SNR) ratio without noise uncertainty for different values Pmd =Pf. 

One additional important parameter in energy detection systems analysis is miss in 

detection Pmd (detector does not detect a signal although it exists). Probabilities Pd and 

Pmd are connected by the equation   

 1md dP P       (7) 

Fig. 2 presents the necessary number of samples (N) as a function of signal-to-noise ratio 

(SNR). The results are presented for equal values of Pf and Pmd. There is no noise uncertainty 

which means that optimum detector threshold value exists independently of SNR. For small 

SNR signal detection is always possible, but the value of N significantly increases. 

In our concrete implementation it is more important to achieve low value of Pmd than 

to achieve low value of Pf. In other words, consequences of miss in detection are more 

severe (RCIED is activated because there is no jamming) than if the detection is false 

(only jamming signal is waste generated). That’s why the results for probability values 

satisfying the condition Pmd<Pf are presented in Fig. 3. 

   

1,E+01

1,E+02

1,E+03

1,E+04

1,E+05

1,E+06

1,E+07

-20 -19 -18 -17 -16 -15 -14 -13 -12 -11 -10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0

SNR(dB)

N

Pf=0.1;Pmd=0.01 Pf=0.01;Pmd=0.001 Pf=0.001;Pmd=0.000001 Pf=0.1;Pmd=0.01;ro=0.25dB

 
Fig. 3 The necessary number of samples (N) when energy detector is applied as a function of 

signal-to-noise (SNR) ratio without noise uncertainty for different values Pmd <Pf. 



  Reliability Analysis of Different RCIED Activation Signal Responsive Jamming Techniques... 465 

 

Noise uncertainty is modelled in such a way that the value ρ-(1/ρ), where it is ρ>1, is 

subtracted from the value SNR in the denominator of equation (5). This means that noise 

power, instead of having power equal to σ
2
 when noise is completely defined, now has 

the value of power between (1/ρ)·σ
2
 and ρ·σ

2
. SNR-wall is presented by the fourth graph 

in Fig. 3. Even for a very small noise uncertainty value ρ=0.25dB or ρ=1.059 when it is 

Pf=0.1 and Pmd=0.01 the value of N tends to infinity for SNR~-9.3dB and below this 

value -9.3dB it is not possible to detect a signal. As a conclusion it may be said that it is 

very important to constantly monitor the noise level and to adjust threshold value 

according to instantaneous noise level and in this way to avoid SNR-wall appearance. 

5. RCIED ACTIVATION SIGNAL DETECTION BY MATCHED FILTER 

The second often implemented technique of spectrum analysis is based on the method of 

matched filters. The main property of such filters is that they are optimum linear filters 

applied for signal detection in white Gaussian noise, meaning that maximum SNR is 

achieved by their implementation [32]. Although this property contributes to easier and 

faster signal detection, the drawback of matched filter implementation is that it is necessary 

to precisely know time characteristics of the signal which has to be detected. Such 

knowledge is possible in some implementation areas, as for example, in cognitive radio 

[32], [33]. But, if considering RCIED activation signal jamming, there is a great variety of 

possible and, in the same time, unpredictable activation techniques. They usually depend on 

the devices which may be easily purchased in some country (region) and easily adapted for 

its malicious function. The number of applied solution types is not great in the analysis 

presented in [34] with the dominant implementation of one type, thus simplifying and 

limiting the necessary number of different matched filters. Nevertheless, application of 

matched filters is not quite suitable for RCIED activation signal detection and the analysis 

of this method has more theoretical than practical significance. 

Similarly to the analysis of energy detector, the necessary number of samples to 

achieve the desired probability of false alarm and probability of successful signal 

detection may be determined on the base of equation from [27]: 

1 1 2
( ( ) ( ))f dQ P Q P

N
SNR

 


     (8) 

Fig. 4 presents the necessary number of samples (N) as a function of signal-to-noise 

ratio (SNR) when matched filter is implemented. The results are obtained by equation (8) 

and are presented for equal values of Pf and Pmd. After that, Fig. 5 presents the 

corresponding results if Pmd<Pf. It is necessary to collect and analyze lower number of 

samples to achieve the same values Pf and Pmd as when energy detector is implemented. 

This is noticeable if graphs from Figs. 2 and 4, as well as graphs from Figs. 3 and 5 are 

mutually compared. 

 



466 M. MILEUSNIĆ, P. PETROVIĆ, V. KOSJER, A. LEBL, B. PAVIĆ  

1,E+00

1,E+01

1,E+02

1,E+03

1,E+04

-20 -19 -18 -17 -16 -15 -14 -13 -12 -11 -10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0

SNR(dB)

N

Pf=Pmd=0.1 Pf=Pmd=0.01 Pf=Pmd=0.001 Pf=Pmd=0.000001
 

Fig. 4 The necessary number of samples (N) when matched filter is applied as a function 

of signal-to-noise (SNR) ratio for different values Pmd =Pf. 

1,E+01

1,E+02

1,E+03

1,E+04

-20 -19 -18 -17 -16 -15 -14 -13 -12 -11 -10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0

SNR(dB)

N

Pf=0.1;Pmd=0.01 Pf=0.01;Pmd=0.001 Pf=0.001;Pmd=0.000001
 

Fig. 5 The necessary number of samples (N) when matched filter is applied as a function 

of signal-to-noise (SNR) for different values Pmd <Pf. 

6. ACTIVE JAMMING USING FREQUENCY SWEEP 

Frequency sweep is often used method of active jamming. It is necessary to linearly 

change signal frequency step by step from its minimum value (f1) to the maximum one (f2) 

in order to realize a sweep. It is a readily implemented jamming method, because a 

significantly smaller power is necessary in relation to wideband jamming based on Additive 

White Gaussian Noise (AWGN) [17] - [19]. Linear frequency change of jamming signal 

frequency is practically approximated by a stepwise change, as it is presented in Fig. 6. 

There are two parameters, besides outmost sweep signal frequencies f1 and f2, which model 

signal frequency change: frequency change step (f∆) and each step time interval duration 

while the same signal frequency is generated (T∆). 



  Reliability Analysis of Different RCIED Activation Signal Responsive Jamming Techniques... 467 

 

f

f1

f2

t

τsw

Wsw

f
∆

T∆

 

Fig. 6 Practical implementation of frequency sweep in the RCIED activation message 

active jamming. 

The total step number in jamming realization may be represented by an equation: 

2 1f f
N

f


      (9) 

while the duration of one total sweep cycle may be represented as: 

2 1
sw

f f
T N T T

f
 




        (10) 

7. COMPARISON OF THE ACTIVE AND RESPONSIVE FFT JAMMING RELIABILITY 

The starting data in our analysis will be the required time to realize sweep jamming of 

RCIED activation signal under the condition that at least once jamming signal frequency 

and RCIED activation message frequency are approximately equal. After that we shall 

determine the total time from the beginning of the analyzed signal sample acquisition 

including FFT calculation time until the start of jamming signal emission on the detected 

RCIED activation signal frequency. In order to achieve comparison requirements of these 

two results, we shall suppose that the number of steps in sweep procedure (N) is equal to 

the number of points where the analyzed signal spectrum is estimated (n). 

There is a number of D/A converters, which may be implemented for jamming signal 

generation. One typical example may be found in [35]. This D/A converter is used in our 

jammer solution [16]. It generates analog signal from the samples, whose maximum 

frequency is pretty high (fs=3.5GHz), thus enabling maximum generated signal frequency 

fg=1.4GHz. 

Let us suppose that RCIED activation jamming is realized by jamming signal generation 

in a frequency band between f1=20MHz and f2≈1.33GHz. We shall further define the 



468 M. MILEUSNIĆ, P. PETROVIĆ, V. KOSJER, A. LEBL, B. PAVIĆ  

frequency change step f∆=20kHz, and the frequency sweep step duration T∆=200ns (very 

short time, a greater time is used in a practical realization - T∆=1μs or more [16]). On the 

base of (9) we obtain the number of steps in a sweep procedure realization N=65536=64K, 

and on the base of (10) sweep procedure duration is Tsw≈13.1ms. 

Let us now have a device for responsive jamming, which collects the analyzed signal 

samples (performs A/D conversion) at the same frequency fs=3.5GHz as the frequency of 

samples for D/A converter [36]. Number of samples that need to be collected, and 

consequently obtained number of the analyzed signal discrete frequency components is the 

same as the number of sweep steps (n=65536=64K). In this way it is achieved that the 

accuracy of jamming signal frequency df according to (2) is the same as the step of sweep 

signal frequency change f∆ according to (9). In the case of responsive jamming, the signal 

sample acquisition time from (1) will be T≈18.7μs, while on the basis of Table 1 the FFT 

calculation time is Tcal=251.7μs. According to our adopted approximation it is also 

Tcomp=251.7μs. Taking into account these three values, the total analysis time on the basis of 

(3) is Tan=522.1μs. 

Comparing two time intervals which are the main indicators of active and responsive 

jamming reliability (Tsw and Tan), it is concluded that responsive jamming on the basis of 

FFT implementation is significantly more reliable (in our example ≈25 times) than active 

jamming by frequency sweep. One additional element which is an advantage of responsive 

jamming is the fact that after the analysis process (Tan) jamming may be performed only on 

the detected activation signal frequency with no time limit. In the case of active jamming, 

signal frequency is only during a short time period T∆ approximately equal to the activation 

signal frequency. Therefore, due to greater jamming time it is also higher the probability 

that jamming is successful when responsive jamming is implemented. 

Fig. 7 presents the results of reliability comparison of responsive jamming based on 

FFT analysis according to data from [24] in relation to active jamming based on 

frequency sweep. When active jamming is considered, jamming time on each frequency 

is adopted to be 200ns. Two extreme cases are chosen for responsive jamming: a) the 

first one, which allows maximum analysis speed (maximum specified processor clock 

0

5

10

15

20

25

30

35

16K 32K 64K 128K 256K 512K 1024K

N FFT

T
s

w
/T

a
n

.

1.4GHz 8 cores 1GHz 1 core
 

Fig. 7 The reliability of responsive jamming on the FFT basis according to [24] in relation 

to active frequency sweep jamming. 



  Reliability Analysis of Different RCIED Activation Signal Responsive Jamming Techniques... 469 

 

frequency 1.4GHz and maximum number of processor cores – 8) and b) the second one, 

which corresponds to the minimum analysis speed (minimum processor clock frequency 

1GHz and minimum number of processor cores – 1). The required number of processor 

cycles for FFT calculation as a function of the number of FFT points is taken according 

to Table 1 from [24]. After that, it is determined the necessary time for FFT analysis in 

some cases. The graph in Fig. 7 indicates higher reliability of responsive jamming on the 

base of FFT in relation to active jamming using frequency sweep in all jamming 

conditions, because the calculated relation of jamming reliability is always between two 

extreme cases presented in Fig. 7 (i.e. this relation is always greater than 1). 

Fig. 8 presents the results of reliability comparison of responsive jamming based on 

FFT analysis according to data from [21] in relation to active jamming based on 

frequency sweep. The processor implemented for FFT calculation operates on lower 

frequencies (between 60MHz and 150MHz) [26] than in the example from Fig. 7. That’s 

why it is adopted that analyzed signal sampling is realized on a significantly lower 

frequency (800MHz) than in the example in Fig. 7. This further means that jamming is 

realized in this case for lower frequencies (till 320MHz). 

0

0,5

1

1,5

2

2,5

8 16 32 64 128 256 512 1024
N FFT

T
s
w

/T
a
n

.

150MHz with HWAFFT 60MHz no HWAFFT 100MHz with HWAFFT 100MHz no HWAFFT

 

Fig. 8 The reliability of responsive jamming on the FFT basis according to [21] in 

relation to active frequency sweep jamming. 

The results in Fig. 8 are presented separately in the case that HWAFFT is used for an 

analysis and when it is avoided. The maximum processing rate in this case is achieved if 

HWAFFT is used together with maximum processor clock frequency (150MHz), while the 

minimum processing rate is if HWAFFT is not used and the processor clock frequency is 

minimum (60MHz). These results are presented by first two vertical graphs for each number 

of frequencies in FFT analysis. Besides these two graphs, the results for mean processor 

clock frequency (100MHz) are presented when HWAFFT is used and when it is not used. 

The required number of processor cycles to calculate FFT for some number of points in 

FFT analysis is determined on the base of Table 3, i.e. Table 4 from [21]. 

Using the analysis the results in Fig. 8 it can be concluded that the application of 

HWAFFT allows also in this case that responsive jamming using FFT may be more 

reliable than the active jamming by frequency sweep (except for the smallest number of 



470 M. MILEUSNIĆ, P. PETROVIĆ, V. KOSJER, A. LEBL, B. PAVIĆ  

analyzed frequencies – 8, which is unlikely to occur in practice). On the contrary, if 

HWAFFT is not used, responsive jamming reliability using FFT becomes lower than the 

reliability of frequency sweep jamming (because the relation Tsw/Tan<1). For the mean 

processor clock frequency (100MHz) and with the HWAFFT implementation for the 

smaller number of points in the analysis, frequency sweep implementation is more 

reliable, while for the greater number of points the analysis on the base of FFT is better. 

8. PERFORMANCES OF ENERGY DETECTOR AND MATCHED FILTER DETECTOR  

ON THE BASE OF MAIN FFT PARAMETERS 

In our further analysis we are going to investigate achievable performances of energy 

detector and matched filter detector on the base of 64K collected (analyzed) signal 

samples, as in the case of FFT analysis. In this way analysis procedure duration is 

comparable for these two methods. The time of sample collection is exactly the same as 

we suppose the same sampling rate in two presented cases. 

Fig. 9 presents the simplified timing diagram of a jammer realized on the base of energy 

detector. The value SNR is relatively low when the whole available frequency bandwidth is 

analyzed at once, such that the necessary number of samples (N) for the analysis is higher 

than 64K. That’s why the complete frequency bandwidth of the analyzed system is separated 

into n distinguished blocks, meaning that bandpass filtering (BPF) is the first step at each of n 

executed block inputs for energy detection (ED). At the output of each block is comparator 

(COMP) and after that (in the block DECISION) is determined whether RCIED activation 

signal is present. The principle structure is the same if energy detector is replaced by matched 

filter (MF), except that designations ED1...EDn in Fig. 9 have to be replaced by MF1...MFn. 

There are two algorithm execution possibilities: 1. sequential processing in each of n blocks 

(Fig. 9a)); 2. parallel processing in n blocks when considering the elapsed time (Fig. 9b)). Of 

course, the combination of these two structures is also possible, but this time scenario is 

beyond the scope of our analysis. If the white noise power in the whole available frequency 

bandwidth is designated by σv
2
, the power at the output of each of n bandpass filters is σv

2
/n, 

because white noise frequency spectrum is considered to be uniform and it is split into n 

equal frequency portions. It means that SNR in the energy detector where RCIED activation 

signal appears is n times higher than without spectrum separation. 

 

BPFn
t

EDn COMPn DECISIONn BPF1 ED1 COMP1 DECISION1...

 
a) 

 

BPFn

t

EDn COMPn DECISIONn

BPF1 ED1 COMP1 DECISION1

..
.

..
.

..
.

..
.

 
b) 

Fig. 9 The timing diagram of the energy detector based jammer: a) sequential processing; 

b) parallel processing 



  Reliability Analysis of Different RCIED Activation Signal Responsive Jamming Techniques... 471 

 

Fig. 10 presents the total number of analyzed samples (N) to successively perform 

energy detection as a function of the number of blocks (frequency spectrum parts) n when 

SNR=-20dB for the whole frequency bandwidth in the case that sequential processing is 

applied as in Fig. 9a). The total number of collected samples is determined as a product of 

the number of sequentially realized blocks and the necessary number of samples for each 

block to achieve the noise power σv
2
/n. The results are presented for three pairs of values Pf 

and Pmd when there is no noise power uncertainty and for one of these parameter pairs when 

the value of noise power uncertainty is ρ=0.25dB. We want to have 64K analyzed samples 

as in the case of FFT based detection and we may determine the necessary number of 

sequential blocks for different Pf and Pmd. For example, when it is Pf=0.1 and Pmd=0.01, it is 

enough to have n=5 sequential blocks when there is no noise power uncertainty. This value 

increases to even n=24 when noise power uncertainty is only ρ=0.25dB. It is expected that 

such a low ρ is very rare, because there are usually a lot of different signals besides white 

noise in the wireless system surrounding. 

 

1,E+03

1,E+04

1,E+05

1,E+06

1,E+07

1 10 100

number of sequential blocks

N

Pf=0.1;Pmd=0.01 Pf=0.01;Pmd=0.001 Pf=0.001;Pmd=0.000001 Pf=0.1;Pmd=0.01;ro=0.25dB
 

Fig. 10 The necessary number of analyzed samples (N) as a function of the number of 

analysis blocks in the case of energy detector implementation when SNR=-20dB 

for the whole frequency bandwidth when sequential processing is used. 

Fig. 11 presents the total number of analyzed samples (N) to perform energy detection as 

a function of the number of blocks (frequency spectrum parts) n when SNR=-20dB  for the 

whole frequency bandwidth in the case that parallel processing is applied as in Fig. 9b). The 

parameters on this figure are the same as for Fig. 10. If there is no noise power uncertainty, it 

is enough to have only 2 parallel blocks and to collect about 64K samples when it is Pf=0.1 

and Pmd=0.01. With the noise power uncertainty we ought to have 14 blocks.   

 



472 M. MILEUSNIĆ, P. PETROVIĆ, V. KOSJER, A. LEBL, B. PAVIĆ  

1,E+01

1,E+02

1,E+03

1,E+04

1,E+05

1,E+06

1,E+07

1 10 100

number of parallel blocks

N

Pf=0.1;Pmd=0.01 Pf=0.01;Pmd=0.001 Pf=0.001;Pmd=0.000001 Pf=0.1;Pmd=0.01;ro=0.25dB
 

Fig. 11 The necessary number of analyzed samples (N) as a function of the number of 

analysis blocks in the case of energy detector implementation when SNR=-20dB 

for the whole frequency bandwidth when parallel processing is used. 

 

Fig. 12 presents the total number of analyzed samples (N) to perform matched filter 

detection as a function of the number of blocks (frequency spectrum parts) n when   SNR=-

40dB for the whole frequency bandwidth in the case that sequential processing is applied as 

in Fig. 9a). The conditions for the graphical presentation are the same as in Fig. 10. In this 

case the number of necessary points for the analysis is independent of the number of applied 

sequential blocks. Such behaviour is the consequence of the fact that SNR is linear factor in 

the denominator of equation (8). It means that algorithm characteristics may not be 

improved by sequential processing for matched filter implementation. Opposite to this, SNR 

appears as the quadratic factor in the denominator of equation (5) for energy detector. Also 

SNR appears under the square root of the numerator in the same equation. Such complex 

dependence causes decreasing of the number of necessary samples when the number of 

analysis blocks increase in the case of energy detector implementation. 

 

1,E+05

1,E+06

1 10 100

number of sequential blocks

N

Pf=0.1;Pmd=0.01 Pf=0.01;Pmd=0.001 Pf=0.001;Pmd=0.000001
 

Fig. 12 The necessary number of analyzed samples (N) as a function of the number of 

analysis blocks in the case of matched filter detector implementation when SNR= 

-40dB for the whole frequency bandwidth when sequential processing is used. 



  Reliability Analysis of Different RCIED Activation Signal Responsive Jamming Techniques... 473 

 

Fig. 13 presents the total number of analyzed samples (N) to perform matched filter 

detection as a function of the number of blocks (frequency spectrum parts) n when   SNR=-

40dB for the whole frequency bandwidth in the case that parallel processing is applied as in 

Fig. 9b). The results are presented for hundred times worse value of SNR than in the case of 

energy detector consideration. This is one more proof of the better matched filter 

performances comparing to energy detector. 

1,E+03

1,E+04

1,E+05

1,E+06

1 10 100

number of parallel blocks

N

Pf=0.1;Pmd=0.01 Pf=0.01;Pmd=0.001 Pf=0.001;Pmd=0.000001
 

Fig. 13 The necessary number of analyzed samples (N) as a function of the number of 

analysis blocks in the case of matched filter detector implementation when SNR= 

-40dB for the whole frequency bandwidth when parallel processing is used. 

Energy detector or matched filter detector as the base of RCIED activation jammer are 

more suitable in the case when there is necessary to intercept only a frequency spectrum 

part. In such a case the number of sequential or parallel blocks on the block-schema from 

Fig. 9 is decreased as a consequence of increased SNR, making a solution practically 

realisable. This situation is presented in [34], where it is demonstrated that even in the area 

of significant military war activities, there is not too high number of implemented RCIED 

activation solution types. Therefore, if energy detector or matched filter detector supervises 

the wideband signal, the number of spectrum parts to which the whole frequency bandwidth 

is separated is relatively low (not more than 100 according to figures 10-13). As a 

consequence, each frequency part for detection is relatively wide and it may be efficiently 

jammed only by frequency limited noise (barrage) jamming. Opposite, FFT gives a lot of 

points where frequency spectrum is estimated (64K in the example in Section 7). That’s 

why jamming may be realized on one or more precisely defined frequencies implementing 

pure sinusoid (spot) jamming.    

The pipeline processing is one possibility to several times increase the analysis rate 

without significantly increasing hardware size. This possibility may be considered first of 

all when FFT method is applied, but also when energy detector or matched filter detector 

are considered. The principle details about pipeline processing in RCIED detection are 

found in [1], [37]. 



474 M. MILEUSNIĆ, P. PETROVIĆ, V. KOSJER, A. LEBL, B. PAVIĆ  

8. CONCLUSIONS 

The results of the calculation presented in this paper have proved that responsive 

jamming may be more reliable than active jamming. This is the first paper contribution. The 

required time for secure jamming signal generation on the frequency of RCIED activation 

message is analyzed as a criterion of jamming reliability. As the result of complete analysis 

based on FFT, the frequency of RCIED activation signal is obtained, and jamming on the 

exactly determined frequency may be initiated. The results are compared for two 

processors, which are specialized for FFT calculation. One of these two processors provide 

that responsive jamming based on FFT implementation is more reliable than active jamming 

using frequency sweep. In this case analysis rate is several times, and even up to several 

tens of times higher when FFT is implemented in the analysis in relation to the frequency 

sweep rate. For the second analyzed processor reliability of responsive jamming depends on 

processor hardware characteristics such as the processor clock frequency and whether 

hardware accelerator (HWAFFT) is applied. If HWAFFT is included in the analysis with 

the higher processor clock frequency, jamming based on FFT analysis is certainly more 

reliable. On the contrary, if HWAFFT is not used with the lower processor clock frequency, 

the speed of FFT analysis may not approach jamming speed realized by frequency sweep. 

The second paper contribution is related to the possibilities of energy detector and 

matched filter detector implementation for RCIED activation responsive jamming. It is 

proved that these jamming types are more suitable for jamming of narrower frequency 

bands, because frequency spectrum is estimated in significantly lower number of points 

than it is the case with FFT analysis. The additional problem for energy detector and 

matched filter detector implementation is a priori unknown shape of RCIED activation 

signal and noise power level. To overcome these problems, these two detectors require 

collecting relatively high number of signal samples. It is explained how this number of 

samples may be reduced while considering the desired probabilities of false detection and 

miss of detection, as well as analysis rate (number of collected samples).   

It can be summarized that the results of comparative analysis presented in this paper 

prove that at up-to-date technological development level, RCIED activation responsive 

jamming may be very reliable and very often even more reliable than active jamming, 

especially when FFT analysis is considered. 

This paper is the extended and enhanced version of the contribution [1]. Comparing 

to [1], completely new sections are 2, 4, 5 and 8. Section 2 presents the principle block-

schema of the analyzed solutions to simplify readers following the text as two new 

methods are described in the paper. These two new methods are introduced in Sections 4 

and 5. The priority in the analysis is given to the determination of necessary number of 

samples to achieve comparable timing performances to the RCIED activation signal 

detector realized by FFT. The main new results for practical jammer realization are in the 

Section 8, where it is analyzed how to choose the most important jammer specification 

parameters: the necessary number of collected samples and the number of sequential or 

parallel blocks to achieve the desired jammer characteristics.      

Acknowledgement: The paper is realized in the framework of the project TR32051, which is 

cofinanced by Ministry of Education, Science and Technological Development of the Republic of Serbia, 

2011-2019. 



  Reliability Analysis of Different RCIED Activation Signal Responsive Jamming Techniques... 475 

 

REFERENCES  

[1] M. Mileusnić, P. Petrović, A. Lebl, B. Pavić, “Comparison of RCIED Activation Responsive and Active 

Jamming Reliability”, In Proceedings of the 6
th
 International Conference IcETRAN 2019, Silver Lake, 

June 3-5, 2019, pp. 988–993. 

[2] P. Petrović and M. Šunjevarić, “Radio Surveillance and Jamming Systems and Techniques“, IN 

Proceedings of EEE Conference: Trends in telecommunications development (organized by Faculty of 

Electrical Engineering and the company Electronics Industry), Belgrade, November 1988, pp. 17.1–

17.22., in Serbian. 

[3] P. Petrović, “Syllabus EW Course: Introductory Course of Electronic Warfare Systems, Techniques and 

Technologies”, Preliminary material for lectures at the Royal Academy in Shrivenham, Research Gate, 

January 2011. 

[4] R. Poisel, Modern Communications Jamming Principles and Techniques. Second Edition, Artech House, 

Boston/London, 2011. 

[5] Security and Defense Technologies (Secintel), “Portable RF Jammer“, http://www.secintel.com/ecom-

prodshow/portable_rf_jammer.html. 

[6] Homeland Security Strategies GB LTD, “VIP-300T Covert IED Jammer – Trunk Mounted Version“, 

http://www.secintel.com/media/pdf/vip300T_Covert_IED_Jammer.pdf. 

[7] Security and Defense Technologies (Secintel), “Reseller Demonstration Kits“, http://www.secintel.com/ecom-

prodshow/stationary_jammer.html. 

[8] Security and Defense Technologies (Secintel), “Backpack Jammer“, http://www.secintel.com/ecom-

prodshow/backpack_jammer.html. 

[9] HSS Development, “Technologies + Solutions“, http://www.secintel.com/media/pdf/hsscatalog.pdf. 

[10] Phantom Technologies LTD., “Our Security Solutions”, http://www.phantom.co.il/ufiles/attaches/ 

RCJ1390LT-H_new.pdf. 

[11] Elisra Electronic Systems ltd, “EJAB Family Electronic Jammers Against Bombs“, http://www.mw-

elisra.com/pdf/EJAB-BrochureLIGHT001.pdf. 

[12] Aselsan, “GERGEDAN – Portable RCIED Jammer System (Vehicle Type)“, http://www.aselsan.com.tr/en-

us/capabilities/electronic-warfare-systems/electronik-support-and-electronic-attack-systems/gergedan-portable-

rcied-jammer-system-%28vehicle-type%29. 

[13] J. Mietzner, P. Nickel, A. Meusling, P. Loos and G. Bauch, “Responsive communications jamming 

against radio-controlled improvised explosive devices”, IEEE Communications Magazine, vol. 50, no. 

10, pp. 38–46, October 2012. 

[14] K. Wilgucki. R. Urban, G. Baranowski, P. Grądzki, P. Grądzki and P. Skarźyński, “Automated protection 

system against RCIED”, In Proc. Military Communication Institute (MCI), 2012. 

[15] P. Petrović, N. Remenski, P. Jovanović, V. Tadić, B. Pavić, M. Mileusnić and B. Mišković, “WRJ 2004 

Wideband Radio Jammer against RCIEDs”, technical solution – Project TR32051, 2011, 

http://www.iritel.com/images/pdf/wrj2004-e.pdf. 

[16] M. Mileusnić, P. Petrović, B. Pavić, V. Marinković-Nedelicki, J. Glišović, A. Lebl and I. Marjanović, 

“The Radio Jammer Against Remote Controlled Improvised Explosive Devices“, In Proceedings of the 

25
th
 Telecommunications Forum (TELFOR), Belgrade, November 21-22

th
 2017, pp. 151–154. 

[17] M. Mileusnić, B. Pavić, V. Marinković-Nedelicki, P. Petrović, D. Mitić and A. Lebl, “Analysis of 

Jamming Successfulness against RCIED Activation“, In Proceedings of the 5
th
 International Conference 

IcETRAN 2018, Palić, June 11-14
th
 2018, pp. 1206–1211. Paper awarded as the best one in the section 

Telecommunications. 

[18] M. Mileusnić, P. Petrović, B. Pavić, V. Marinković-Nedelicki, V. Matić and A. Lebl, “Jamming of 

MPSK Modulated Messages for RCIED Activation”, In Proceedings of the 8
th
 International Scientific 

Conference on Defensive Technologies OTEH 2018, Belgrade, 11-12
th
 October 2018, pp. 380–385. 

[19] M. Mileusnić, B. Pavić, V. Marinković-Nedelicki, P. Petrović, D. Mitić and A. Lebl, “Analysis of jamming 

successfulness against RCIED activation with the emphasis on sweep jamming,“ Facta Universitatis, Series 

Electronics and Energetics, Vol. 32, No. 2, June 2019, https://doi.org/10.2298/FUEE1902211M, ISSN: 0353-

3670, pp. 211-229.,  the extended and revised version of the paper from the IcETRAN 2018. 
[20] R. R. Jaglan, S. Sarowa, R. Mustafa, S. Agrawal, N. Kumar, “Comparative Study of Single-user Spectrum 

Sensing Techniques in Cognitive Radio Networks”, In Proceedigs of the Second International Symposium on 

Computer Vision and the Internet (VisionNet’15), Procedia Computer Science 58, 2015, pp. 121–128. 

[21] M. McKeown, “Texas Instruments: FFT Implementation on the TMS320VC5505, TMS320C5505, and 

TMS320C5515 DSPs“, Application Report SPRABB, June 2010 – Revised January 2013, pp. 1-28. 

http://www.secintel.com/ecom-prodshow/portable_rf_jammer.html
http://www.secintel.com/ecom-prodshow/portable_rf_jammer.html
http://www.secintel.com/media/pdf/vip300T_Covert_IED_Jammer.pdf
http://www.secintel.com/ecom-prodshow/stationary_jammer.html
http://www.secintel.com/ecom-prodshow/stationary_jammer.html
http://www.secintel.com/ecom-prodshow/backpack_jammer.html
http://www.secintel.com/ecom-prodshow/backpack_jammer.html
http://www.secintel.com/media/pdf/hsscatalog.pdf
http://www.phantom.co.il/ufiles/attaches/%20RCJ1390LT-H_new.pdf
http://www.phantom.co.il/ufiles/attaches/%20RCJ1390LT-H_new.pdf
http://www.mw-elisra.com/pdf/EJAB-BrochureLIGHT001.pdf
http://www.mw-elisra.com/pdf/EJAB-BrochureLIGHT001.pdf
http://www.aselsan.com.tr/en-us/capabilities/electronic-warfare-systems/electronik-support-and-electronic-attack-systems/gergedan-portable-rcied-jammer-system-%28vehicle-type%29
http://www.aselsan.com.tr/en-us/capabilities/electronic-warfare-systems/electronik-support-and-electronic-attack-systems/gergedan-portable-rcied-jammer-system-%28vehicle-type%29
http://www.aselsan.com.tr/en-us/capabilities/electronic-warfare-systems/electronik-support-and-electronic-attack-systems/gergedan-portable-rcied-jammer-system-%28vehicle-type%29
http://www.iritel.com/images/pdf/wrj2004-e.pdf
https://doi.org/10.2298/FUEE1902211M


476 M. MILEUSNIĆ, P. PETROVIĆ, V. KOSJER, A. LEBL, B. PAVIĆ  

[22] E. Çetin, R. C. S. Morling and I. Kale, “An Integrated 256-point Complex FFT Processor for Real-time 

Spectrum Analysis and Measurement”, In Proceedings of the IEEE Proceedings of Instrumentation and 

Measurement Technology Conference, May 1997, vol. 1, pp. 96–101. 

[23] “Pipelined FFT/IFFT 256 points (Fast Fourier Transform) IP Core User Manual“, Unicore Systems Ltd., pp. 1-

17., https://opencores.org/websvn/filedetails?repname=pipelined_fft_256&path=%2Fpipelined _fft_256% 

2Ftrunk%2FDOC%2Ffft256_um.pdf. 

[24] X. Li, E. Blinka, “Very large FFT for TMS320C6678 processors“, Texas Instruments, 2015., pp. 1-6. 

[25] Texas Instruments, “Multicore Fixed and Floating-Point Digital Signal Processor“, SPRS691 – 

November 2010 – Revised March 2014., pp. 1-242. 

[26] Texas Instruments, “TMS320C5505 Fixed-Point Digital Signal Processor“, SPRS660F – August 2010 – 

Revised September 2013., pp. 1-157. 

[27] T. Trump and I. Müürsepp, “Detection Speed of Responsive Communication Jamming Detectors, Recent 

Advances in Telecommunications and Circuits”, 2nd International Conference on Circuits, Systems, 

Communications, Computers and Applications, Dubrovnik, June 2013, pp. 149-154. 

[28] S. Atapattu, C. Tellambura and H. Jiang, Energy Detection for Spectrum Sensing in Cognitive Radio. 

Springer, 2014, Chapters 1-2, ISSN 2191-5768, ISBN 978-1-4939-0493-8, DOI 10.1007/978-1-4939-

0494-5, pp. 1-26. 

[29] S. Bahamou and A. Nafkha, “Noise Uncertainty Analysis of Energy Detector: Bounded and Unbounded 

Approximation Relationship”, In Proceedings of the 21
st
 European Processing Conference (EUSIPCO 

2013), 9-13. September 2013, Marrakech, Morocco, pp. 1–4. 

[30] R. Tandra and A. Sahai, “SNR walls for signal detection”, IEEE Journal on Selected Topics in Signal 

Processing, vol. 2, no. 1, pp. 1–30, 2008. 

[31] M. M. Tanatwy, “Responsive Communication Jamming Detector with Noise Power fluctuation using 

Cognitive Radio”, International Journal of Innovative Research in Computer and Communication 

Engineering, vol. 2, no. 10, pp. 5967–5973, October 2014. 

[32] S. Dannana, B. Prasad Chapa, G. S. Rao, “Spectrum Sensing Using Matched Filter Detection”, Intelligent 

Engineering Informatics: Proceedings of the 6
th
 International Conference on Frontiers of Intelligent 

Computing: Theory and Applications (FICTA), 2017, Bhubaneswar, Odisha, Springer Verlag, Singapore,  

pp. 497-503, ISBN: 9811075654.  

[33] F. Salahdine, H. E. Ghazi, N. Kaabouch, W. F. Fihri, “Matched Filter Detection with Dynamic Threshold 

for Cognitive Radio Networks”, In Proceedings of the 2015 International Conference on Wireless 

Networks and Mobile Communications (WINCOM), 23
th
 October 2015, France.  

[34] G. Attila, “The Radio Controlled Improvised Explosive Device (RCIED) threat in Afghanistan, AARMS, 

vol. 12, no. 1, 2013, pp. 9–23.     

[35] Analog Devices, “AD9914: 3.5 GSPS Direct Digital Synthesizer with 12-Bit DAC“, Rev. F, February 

2017., pp. 1-46. 

[36] Texas Instruments, “ADC12J4000, 12-Bit, 4GSPS ADC With Integrated DDC“, SLAS989D – January 

2014 – Revised October 2017., pp. 1-96. 

[37] L. Karlsson, “Method, System and Apparatus for Maximizing a Jammer’s Time-on-Target and Power-on-

Target”, United States Patent Application, Publication No. US 2006/0164283 A1, 27. July 2006. 

 

 

 

 

https://opencores.org/websvn/filedetails?repname=pipelined_fft_256&path=%2Fpipelined%20_fft_256%2Ftrunk%2FDOC%2Ffft256_um.pdf
https://opencores.org/websvn/filedetails?repname=pipelined_fft_256&path=%2Fpipelined%20_fft_256%2Ftrunk%2FDOC%2Ffft256_um.pdf