Design & Implementation Of Stand Alone High Gain

Jammer system Against UAVS Using Hlical Antenna

Anas Mohamed Adel Mahmoud Mohamed Mahmoud Talaat Farag d abdalrady Ebrahim
Misr University For Science and Misr University For Science and Misr University For Science and Misr University For Science and
Technology Technology Technology Technology
Department of electronics and Department of electronics and Department of electronics and Department of electronics and
communication engineering communication engineering communication engineering communication engineering
6 October, Giza 6 October, Giza 6 October, Giza 6 October, Giza
elmolla4@gmail.com adelsedeek26102000@gmail.com 91621@must.edu.eg 91636@must.edu.eg
Mohame

Passant Nagi Mourice Rana Adel Mohamed Merna Baher Badry Rawan Mahmoud
Misr University For Science and Misr University for Science and Misr University for Science and Misr University for Science and
Technology Technology Technology Technology
Department of electronics and Department of electronics and Department of electronics and Department of electronics and
communication engineering communication engineering communication engineering communication engineering
6 October, Giza 6 October, Giza 6 October, Giza 6 October, Giza
basant.nagi@gmail.com ranaadel297@gmail.com 91400@must.edu.eg 91630@must.edu.eg

Abstract— This research presents a novel standalone high- paper aims not only to provide them with autonomous
gain jammer system designed to disrupt unmanned aerial jamming systems. contribute to the field of drone
vehicles (UAVs) operating on GNSS frequencies. The system security, but also promote discussions on the
utilizes four strategically placed helix antennas to achieve broad
responsible implementation and regulation of jamming
frequency coverage and employs a YOLO-based tracking system
for accurate jamming. To ensure continuous operation, a solar-
systems within the GNSS frequency range. By
powered standby system provides reliable backup for the exploring the Capabilities of helical antennas and
tracking unit. Experimental evaluations demonstrate the integrating.
effectiveness of the system in disrupting UAVs, highlighting its
potential to enhance security in sensitive areas and protect II. RELATED WORK Selecting a Template
critical infrastructure. This research contributes to the field of The jamming effectiveness of commercially available
UAV countermeasures by offering a practical and efficient low-cost jammers against UAVs has been studied by J.
solution for mitigating the growing threat posed by unauthorized Farlik et al. (2016) where it is concluded that jamming
drone activity. of the GNSS signal reception can be achieved from a
Keywords—GNSS, UAV, Jamming, AI, YOLO fair distance (a couple of hundred meters from the
UAV). On the other hand, the study finds that jamming
I. INTRODUCTION the remote control signals with the considered jammers
Various types of drones exist on the consumer market is not effective, even when the jammer is much closer
spanning from pieces dedicated to sport or to the UAV than the pilot with the remote controller
entertainment to high-class devices capable of carrying and we have other modules with wide bandwidth like
out complex tasks. Besides using drones along with The AH-1117-13 uses extended stubs to provide the
positive legislation, even drones with basic greater gain while maintaining the wider beam-width
functionality can potentially be used for malicious and designed to provide wideband directional
activities in scenarios that are not socially tolerable or transmission/reception of radio signals from 1100-1700
harmful. For example, a fundamental drone can be used MHz bands including GPS L1, GPS L2, L5, Lo bands.
by the paparazzi to invade celebrities’ privacy or by These low-profile antennas use Circular Polarization
criminals and terrorists to compromise the security of a Antenna Technology - which delivers better penetration
potential target. Therefore, in parallel with the through obstruction and interference. Both Right Hand
development of drones, various anti-drone systems Circular Polarized (RHCP) and Left Hand Circular
have been developed as well [1]. Polarized (LHCP) helical antenna models are available
[2]. Since there are various sorts of drones available in
The primary objective of this research paper is to the consumer market, ranging from sports and
design, develop, and evaluate a standalone jamming entertainment to high-end machines capable of
system capable of effectively countering UAVs within performing sophisticated tasks. Alongside the
the GNSS frequency range using helical antennas. This

XXX-X-XXXX-XXXX-X/XX/$XX.00 ©20XX IEEE

development of drones, different anti-drone devices second. A smaller version, Fast YOLO, achieves an
have also been created. astounding 155 frames per second while maintaining
Our study explores a scenario when a radio-controlled double the mAP (mean average precision) of other real-
drone enters an active zone of an anti-drone system. time detectors.
Sarah Kentsch, et al.(2020) studied and quantified [6] “A novel finetuned YOLOv6 transfer learning
issues related to the use of DL with their own model for real-time object detection” proposes a
UAVacquired images in forestry applications such as: pruning and fine-tuning algorithm along with transfer
the effect of Transfer Learning (TL) and the Deep learning to enhance the efficiency of the YOLOv6
Learning architecture .They used two different Deep model in terms of detection accuracy and inference
Learning architectures (ResNet50 and UUNet), two in- speed [7]. “Real-time object detection and
house datasets (winter and coastal forest) and focus on segmentation technology: an analysis of the YOLO
two separate problem formalizations (Multi- algorithm” explores the YOLO algorithm’s
Label)Patch or MLP classification and semantic development and its significance in real-time object
segmentation). detection and segmentation. [8]
They demonstrated the applicability of the patch-based
III. Materials and Methods
framework with the ResNet50 architecture in a
different and complex example: Detection of the A. Drone Model
invasive broadleaf deciduous black locust (Robinia Insights into developing effective countermeasures
pseudo acacia) in an evergreen coniferous black pine against unauthorized drone activities while ensuring
(Pinus thunbergii) coastal forest typical of Japan. In this that legitimate GNSS services are maintained. Galileo
case, we detected images containing the invasive is already present in more than 30% of the receivers’
species with 75% True Positives (TP) and 9% False models used for drone applications The drone market is
Positives (FP) while the detection of native trees was booming and is set to outstrip any other GNSS user
95% TP and 10% FP. [3] On the other hand, Min-Fan base in aviation and open up new business
Ricky Lee et al. (2023) Proposed a method to avoid opportunities for application developers, according to a
several drawbacks which are uncertainty as the White Paper on European Global Navigation Satellite
variation of occlusion, illumination, season, and Systems (EGNSS) for drones’ operations, produced by
viewpoints. the European GNSS Agency (GSA). GNSS is not an
They applied various complex and uncertain option for drones anymore, but a necessary asset.
environment scenarios are applied to evaluate the GNSS is essential for the safe and reliable navigation of
proposed system via the deep learning model’s drones, and GNSS receivers are implemented on almost
prediction metrics (accuracy, precision, recall, P-R all new commercial drones as standard. With increasing
curve, F1 score). The tracking rate based on the demand for beyond visual line of sight (BVLOS)
Siamese Region Proposal Network Algorithm is up to operations GNSS, possibly with various augmentations,
180 FPS [4]. Tanvir is the most obvious choice of technology for
Ahmad et al. (2020) Proposed a modified YOLOv1- navigation, although it is not the only one. EGNSS for
based neural network is proposed for object detection. added accuracy Given the additional accuracy that
The new neural network model has been improved in Galileo offers, it is no surprise that Galileo is already
the following ways. Firstly, modification is made to the present in more than 30% of the receivers used for
loss function of the YOLOv1 network. drone applications, and many of them also implement
'The improved model replaces the margin style with the EGNOS corrections to increase accuracy. The GSA
proportion style. Compared to the old loss function, the White paper provides an overview of the added value
new one is more flexible and more reasonable in of EGNOS and Galileo for current and emerging
optimizing the network error. Secondly, a spatial operations, as well as for future U-Space services.[9]
pyramid pooling layer is added; thirdly, an inception B. Jammer Model System
model with a convolution kernel of 1 1 is added, which Design of Helical Antenna and Specifications
reduces the number of weight parameters of the layers. The helix antenna can be considered as a spring with N
Extensive experiments on Pascal VOC datasets in
turns with a reflector. The circumference (C) of a turn is
2007/2012 showed that the proposed method achieved
better performance [5]. approximately one wavelength (λ), and the distance (d)
"You Only Look Once: Unified, Real-Time Object between the turns is approx. 0.25C. The size of the
Detection” by Joseph Redmon, Santosh Divvala, Ross reflector (R) is equal to C or λ and can be a circle or a
Girshick, and Ali Farhadi. This paper introduces square. The design yields circular polarization (CP),
YOLO, a novel approach to object detection that which can be either 'right hand' or 'left hand' (RHCP or
frames it as a regression problem. The unified LHCP respectively), depending upon how the helix is
architecture is extremely fast, with the base YOLO wound. To have maximum transfer of energy, both ends
model processing images in real-time at 45 frames per

2|Page
of the link must use the same polarization, unless you use
a (passive) reflector in the radio path.
The gain (G) of the antenna, relative to an isotope (dBi), can
be estimated by:
1) G = 11.8 + 10 * log {(C/ λ) ^2 * N * d} dBi
Helical Antenna Gain (G) = (10.8+10 x log (number of turns x
space between coils))
Diameter (D) = wavelength/3.143
Space between coils (S) = Space between coils x wavelengths
Length of wire (L) = Number of turns x √ (wavelength^2 +
Space between coils^25
Half Power Beam Width of antenna = 52 / (√ (Number of
turns x space between coils))
Beam Width First Nulls of antenna = 115 / (√ (Number of
turns x Space between coils))
Aperture = (10(Antenna Gain / 10) x wavelength^2) / (4 *
3.14)

The Antennas Design & Calculations

1) For 1575.42 MHZ Helix

Fig.2
.

Return loss 1-2 GHZ

Parameters of the matching transformer 140/50

Ω
 Triangle length L: 133 mm
 Triangle Height W: 40 mm

Fig.3. The SWR result 1195.192MHZ Antenna.

2. The Results of the 1195.192MHZ Helix

3. The Results of the 1455.1MHZ Helix

3|Page
 Fig.7. The Smith chart of 1606 MHZ Antenna

Fig.5.
Fig.8. Return loss 1-2 GHZ.
Return loss 1-2 GHZ.

Fig.9. The SWR result 1606 MHZ Antenna

Method Of Connection:
We expressed the method of connection as show in figure 14.
1. Raspberry Pi 4B is Connected to Arduino uno and
Camera via USB.
2. Arduino Uno is Connected to Motor Driver (TB6600)
Which is responsible of controlling stepper motor.
3. Motor Driver is Connected to Power Supply of 24V and
10A. Stepper Motor (Nema 23), Which has a holding
torque of 15 kg.cm, is Also Connected to the Driver To
ensure the accuracy and functionality of the motor.

Fig.6. Return loss 1-2 GHZ.

4. The Results of the 1606 MHZ Helix

4|Page
Fig.10. the connection of tracker system
 SYSTEM FLOWCHART Various evaluation metrics such as precision, recall, and mean
Average Precision (mAP) are calculated to quantify the model's
performance.
Results:

Fig.11. system flowchart

D) YOLO MODEL
The curve is a precision-recall curve for the model. It shows the
Object detection, a crucial task in computer vision, involves trade-off between the model's precision (the ability to correctly
the identification and localization of objects within an image or identify a drone) and its recall (the ability to find all the drones
video. It serves as a foundation for various applications, in an image).
ranging from autonomous vehicles and robotics to security
systems and augmented reality. In recent years, the demand for The model has a high precision of 0.962 when the recall is
real-time and accurate object detection has grown 0.904. This means that for 90.4% of the drones in the image
exponentially, necessitating the development of novel that the model detects, it is correct 96.2% of the time.
algorithms capable of meeting these requirements. The mAP@0.5 value is 0.933. This stands for "mean Average
The You Only Look Once (YOLO) algorithm has emerged Precision at an Intersection over Union (IoU) threshold of 0.5".
as a groundbreaking solution to real-time object detection. IoU is a metric that measures how well a predicted bounding
Unlike traditional multi- step approaches that rely on region box overlaps with the actual drone in the image. An IoU of 0.5
proposal networks (RPNs) and subsequent refinement, YOLO means that the bounding box must overlap with the drone by at
follows a unified pipeline, processing the entire image at once Fig.12. precision-recall curve
to predict bounding boxes and class probabilities [10]. This least 50% for it to be counted as a correct detection. The
unique design philosophy offers remarkable speed advantages, mAP@0.5 value of 0.933 means that the model has an average
making it highly suitable for real-time applications.

Detection model:
The YOLOv8 model is a state-of- the-art object detection
algorithm that detects drones of different shapes and sizes. It
outperforms previous YOLO versions in terms of accuracy and
speed. The model is built using PyTorch and integrated with
OpenCV for image processing and used to send control signals
to these devices, allowing for seamless integration between the
detection algorithm and the physical jamming mechanism [11].

Dataset preparation and training: precision of 0.933 when the IoU threshold is set to 0.5.
We collected a dataset of 4030 images of different drones’ Fig.13. recall-confidence curve.
shapes and then split up into 70% training set and 30%
validation set. The curve is a recall-confidence curve for the model. It shows
the trade- off between the model's recall (the ability to find all
Then the model is trained through 100 epochs. During
the drones in an image) and its confidence (how certain the
training, the model learns to identify and localize drones in
model is that a detection is actually a drone). The model has
images accurately. The loss function is minimized through
high recall (0.904) at a reasonable confidence level (0.962).
backpropagation, resulting in improved detection performance
over time. This means that the model can find a large proportion of the
drones in an image (90.4%) while still being fairly certain
(96.2%) that they are actually drones.

5|Page
 The mAP@0.5 value is 0.933. This stands for "mean 0.97: This value represents the True Positive Rate (TPR),
Average Precision at an Intersection over Union (IoU) also known as recall, which means 97% of the actual drones in
threshold of 0.5". IoU is a metric that measures how well a the images were correctly identified by the model.
predicted bounding box overlaps with the actual drone in the
image. An IoU of 0.5 means that the bounding box must
overlap with the drone by at least 50% for it to be counted as a 0.03: This value represents the False Positive Rate (FPR),
correct detection. The mAP@0.5 value of 0.933 means that the which means 3% of the time, the model identified something as
model has an average precision of 0.933 when the IoU a drone when it wasn't actually a drone.
0.95: This value represents the True Negative Rate (TNR),
which means 95% of the time, the model correctly identified
non-drones as non-drones.
0.05: This value represents the False Negative Rate (FNR),
which means 5% of the time, the model missed a drone and
didn't identify it.

Fig.14. precision-confidence curve

threshold is set to 0.5.

The curve is a precision-confidence curve. It shows the trade-

off between the model's precision (the fraction of correct
detections among all detections) and its confidence (how
certain the model is that a detection is actually a drone).
The model has high precision (1.00) at a confidence level of Fig.16. precision-recall curves
0.906. This means that when the model is very confident mAP@0.5: 0.933. This means the model has an average
(confidence ≥ 0.906) in a detection, it is almost always correct precision of 0.933 when the IoU threshold is set to 0.5.
(precision = 1.00).
Precision at recall of 0.904: 0.962.
The mAP@0.5 value is not provided in the image.
However, we can see that the average precision across all This means that when the model recalls 90.4% of the drones
confidence levels is close to 1.0, which suggests that the model (i.e., finds 90.4% of all actual drones), it has a precision of
is very good at making accurate detections. 96.2% (meaning 96.2% of its detections are correct).
The image shows a confusion matrix for the model. mAP@0.95: 0.600. This means the model has an average
precision of 0.600 when the stricter IoU threshold of 95% is
applied.
Precision at recall of 0.5: 0.745. This means that when the
model recalls 50% of the drones with the stricter IoU threshold,
it has a precision of 74.5%.
The model seems to be performing well at detecting drones in
images, with a high recall (97%) and a reasonable precision
(around 0.933 at an IoU of 0.5). However, there is always room
for improvement, and future work could focus on reducing the
false positive rate (3%) and the false negative rate (5%).

E) SOLAR PANEL
We used solar panels as a backup system instead of a power
source. Solar panels consist of a solar cell, which is an
Confusion Matrix: electronic device that converts sunlight directly into electricity.
Fig.15. confusion matrix Light shining on a solar cell produces a current and voltage to
generate electrical energy. This process requires firstly a
material in which the absorption of light raises the electron to a

6|Page
higher energy state, and secondly, the movement of this higher- overload protection and short circuit protection and
energy electron from the solar cell to an external circuit. The overcharging protection.
electron then dissipates its energy in the external circuit and
3. Solar Battery
returns to the solar cell. A variety of materials and processes
can potentially meet the requirements for PV power Batteries that store the excess electricity generated by your
conversion, but in practice, almost all PV power conversion solar panels, which can later be as a power source to generate
processes use semiconducting materials in the form of a p-n electricity.
junction. Regarding the development of sustainable energy
such as solar energy, in this article we will study the types of The economic ones are of the type with lead-acid and most
solar cells and their applications.[12] of them are 12 volts or 24 volts. To deal with it, we need to
know at least two out of three variables: electrical voltage,
STUDY measured in volts, current, measured in amperes, and power,
measured in watts. Battery capacity is indicated in Amps-Hours
The main components of solar panels are silicon solar panels
[15]
consisting of union silicon cells that produce electrical energy.
Battery used in our project is 12 V and 7 Ah (Battery can
HOW DOES IT WORK
deliver up to 7 Amps for an hour or 1 Amp for 7 Hours) and
Solar panels convert solar energy into electrical energy about max Charging Current 2 A
photovoltaic cells. The Word "Photovoltaic "Indicates for the
Connection Between Light "Photon” and Electricity “Volt”
Photovoltaic Effect Provides a DC Current so it Produces a DC
Power Solar photovoltaic panels capture sunlight causing the
electrons in the silicon cells to be related to become direct
Tracker System

Figure 20 Block diagram of connection

Fig.18. Solar system with tracking system

IV.CONCLUSION
In conclusion, this research successfully developed and
implemented a standalone high-gain jammer system for
current. [13] disrupting unmanned aerial vehicles (UAVs) operating on
GNSS frequencies. The system achieved its objectives by
Fig.17. Working principle of solar panel
utilizing strategically placed helix antennas, a YOLO-based
SOLAR PANEL CONSISTS OF: tracking system for improved accuracy, and a reliable solar-
powered standby system. Experimental evaluations
1. Solar cell demonstrated the effectiveness of the system in disrupting
As shown in fig.20 Solar cells in our project are polycrystalline UAVs, highlighting its potential to enhance security in
types with 20-watt maximum power and 18-volt operating sensitive areas and protect critical infrastructure. This research
voltage and 30-volt max voltage and 1.11 max current. contributes to the field of UAV countermeasures by offering a
practical, efficient, and reliable solution for mitigating the
2. Solar Charge Controller growing threat posed by unauthorized drone activity. Our
Also known as a charge regulator, is a voltage and/or current project consists of two systems first one is a tracker system
regulator that prevents batteries from becoming overcharged. It with camera look in random direction once it’s see UAV it’s
regulates the voltage and current going to the battery from the give the information to raspberry pi and it’s give command to
solar panels. Because most "12 volt" panels produce around 16 Arduino to start motor motion in the same direction of camera
to 20 volts, the batteries will be harmed by overcharging if because we have jammer system consist of directive antennas
there is no regulation. To reach full charge, most batteries so when motor moves the jammer system in the direction of
require between 14 and 14.5 volts. drone jamming Figure 19 The system is ready for jamming.
accuracy will be higher than without tracking system In this
With low-maintenance or trickle-charge panels, such as those picture we have mobile phone, and we open an drone figure on
ranging from 1 to 5 watts, a charge controller is usually it and we move it right and left when the camera see this figure
unnecessary. If the panel produces less than 2 watts per 50 it’s sending information to raspberry pi to send a command to
battery amp-hours, you don't need one.[14] Charger Controller Arduino to move the motor then jammer system will move in
(PWM) in our project is 12 to 24 V and 5 A and can be used to the direction of drone fig.20

7|Page
 VI.REFERENCES
[1] Tomislav Milošević (2022), Jamming a Drone
- EM Simulation of Simple
EW and EW Countermeasures Scenarios, PROCEEDINGS, IX
INTERNATIONAL CONFERENCE
IcETRAN, Novi Pazar, Serbia
[2] J. Farlik, M. Kratky, and J. Casar, “Detectability and jamming of
small UAVs by commercially available low-cost means,” in Communications
(COMM), 2016 International Conference on. IEEE, 2016, pp. 327–330.
[3] Kentsch, S., Caceres, M. L. L., Serrano, D., Roure, F., & Díez, Y.
(2020). Computer vision and deep learning techniques for the analysis of
Drone-Acquired Forest Images, a transfer learning study. Remote Sensing,
12(8), 1287. https://doi.org/10.3390/rs12081287
[4] Lee, M. R., & Chen, Y. (2023). Artificial intelligence-based object
detection and tracking for a small underwater robot. Processes, 11(2), 312.
https://doi.org/10.3390/pr11020312
[5] Ahmad, T., Ma, Y., Yahya, M., Ahmad, B., Nazir, S., & Haq, A. U.
(2020). Object Detection through
Modified YOLO Neural Network. Scientific
Programming, 2020, 1–10. https://doi.org/10.1155/2020/8403262
[6] Joseph Redmon, Santosh Divvala, Ross Girshick, Ali Farhadi;
Proceedings of the IEEE Conference on Computer Vision and Pattern
Recognition (CVPR), 2016, pp. 779-788
Fig.19. Our system [7] Gupta, C., Gill, N.S., Gulia, P. et al. A novel finetuned YOLOv6
transfer learning model for real-time object detection. J Real-Time Image Proc
20, 42 (2023). https://doi.org/10.1007/s11554-023-01299-3 [8] Kang, C.H.,
Kim, S.Y. Real-time object detection and segmentation technology: an analysis
of the YOLO algorithm. JMST Adv. 5, 69–76 (2023).
https://doi.org/10.1007/s42791-023-00049-7
[9] European GNSS Agency, (2019). European Global Navigation
Satellite Systems (EGNSS) for drones’ operations: white paper, Publications
Office. https://data.europa.eu/doi/10.2878/52219
[10] Terven, J. and Cordova-Esparza, D. (2023). A Comprehensive
Review of YOLO: From YOLOv1 to YOLOv8 and Beyond, MACHINE
LEARNING AND KNOWLEDGE EXTRACTION.
[11] Reis, D. J., Kupec, J., Hong, J., & Daoudi, A. (2023). Real-Time
Flying Object Detection with
YOLOv8. arXiv (Cornell University).
https://doi.org/10.48550/arxiv.2305.09972
[12] Fahrenburch, A. and Bube, R. 1983, Fundamentals of solar cells,
Academic Press, New York.
[13] Wohlgemuth JH, Narayanan S. Buried contact concentrator solar
cells. Twenty Second IEEE
Photovoltaic Specialists Conference
[14] Solar Power & Battery Systems for Beginners by Michael Holt
Fig.20. The drone moved in drone direction. covers the basics of solar power systems, including charge controllers.
[15] Off Grid Solar: A handbook for Photovoltaics with Lead-Acid or
Lithium-Ion batteries by Dandelion.
V.ACKNOWLEDGEMENT
We would like to express our sincere gratitude and appreciation
to Professor Dr. Mamdouh Gouda & Dr. Ashraf Samy for his
invaluable guidance and support throughout the graduation
project.

8|Page