=Paper=
{{Paper
|id=Vol-3676/BISEC_paper_6
|storemode=property
|title=Energy-Efficient Routing in UAVs Supported Perimeter Security Networks
|pdfUrl=https://ceur-ws.org/Vol-3676/short_06.pdf
|volume=Vol-3676
|authors=Alexander K. Alexandrov,Anastass N. Madzharov
|dblpUrl=https://dblp.org/rec/conf/bisec/AlexandrovM23
}}
==Energy-Efficient Routing in UAVs Supported Perimeter Security Networks==
https://ceur-ws.org/Vol-3676/short_06.pdf
Energy-Efficient Routing in UAVs Supported
Perimeter Security Networks
Alexander K. Alexandrov 1,* , Anastass N. Madzharov 1
1
Institute of Robotics, Bulgarian Academy of Sciences, Acad. G. Bonchev str., 1113 Sofia, Bulgaria
Abstract
UAV-assisted ground and underwater perimeter security sensor networks represent a sophisticated integration of aerial,
ground, and underwater technologies for surveillance and security purposes. This system combines Unmanned Aerial Vehicles
(UAVs) with underwater sensors to monitor and protect strategic areas like harbors, offshore installations, and coastal facilities.
Unmanned Aerial Vehicles (UAVs) have become pivotal in modern surveillance and security operations. Their versatility,
mobility, and technological adaptability make them ideal for perimeter security systems. This study examines the integration
of group of UAVs into perimeter security, evaluating their effectiveness, operational frameworks, technological advancements,
and potential future developments. We analyze and implement a PSO (Particle Swarm Optimization) algorithm, related to
group of UAVs trajectory optimization, review case studies, and identify key considerations for effective development.
Keywords
UAV, PSO, sensor network, perimeter security
1. Introduction sensors is analyzed, processed, and fused to form a com-
prehensive operational picture. Control center assesses
UAV-assisted underwater perimeter security sensor net- potential threats based on the gathered information and
works represent a cutting-edge blend of aerial and mar- coordinates appropriate responses.
itime technologies, designed to enhance the security of One of the challenges in the UAVs assisted underwater
critical aquatic areas. This integration of Unmanned perimeter security sensor networks is the energy man-
Aerial Vehicles (UAVs) and underwater sensors provides agement. Both the UAVs and underwater sensors must
a robust solution for monitoring and safeguarding sen- efficiently manage their power to ensure prolonged oper-
sitive zones like naval bases, coastal areas, ports, and ational capabilities. The present study focuses on energy
offshore installations. management, especially in energy-efficient and reliable
The key components in the UAV-assisted underwater routing of groups of UAVs. The UAVs energy-efficient
perimeter security sensor networks are the sensors, UAVs, routing is a multifaceted challenge that involves optimiz-
and the control center. ing the flight paths and operational strategies of UAVs.
Underwater sensors typically include acoustic sensors The objective is to maintain vigilant monitoring and
(such as sonars), geophones, hydrophones for detecting rapid response capabilities while minimizing energy con-
sound under water, and magnetic anomaly detectors for sumption, which is critical for the longevity and effec-
identifying metallic objects. These sensors continuously tiveness of the UAVs in defense operations. The aim is
scan underwater environments to detect and track po- to create routes and operational patterns that minimize
tential threats, like submarines, divers, or unmanned energy usage while ensuring comprehensive security
underwater vehicles (UUVs). coverage.
UAVs provide real-time aerial surveillance, signifi- Altitude and speed optimization in UAV-supported un-
cantly extending the range of observation beyond the derwater perimeter security sensor networks is a critical
immediate perimeter. They act as a vital link between the aspect of ensuring energy-efficient routing and effective
underwater sensors and the control center, especially im- operation. The right balance of altitude and speed di-
portant in deep-water areas where direct communication rectly impacts the UAVs’ energy consumption, coverage
is difficult. area, sensor effectiveness, and response times.
Control center provides data processing and decision
making. Here the data from both UAVs and underwater
1.1. Altitude optimization
BISEC’23: 14th International Conference on Business Information Higher altitudes can offer less air resistance, but the ben-
Security, November 24, 2023, Niš, Serbia efit must be balanced against increased energy require-
*
Corresponding author. ments for climbing and maintaining altitude. Higher
$ akalexandrov@ir.bas.bg (A. K. A. ); a.madzharov@ir.bas.bg altitudes may increase the coverage area but could re-
(A. N. M. )
© 2024 Copyright for this paper by its authors. Use permitted under Creative Commons License duce the detail or accuracy of sensor data. The right
CEUR
Workshop
Proceedings
http://ceur-ws.org
ISSN 1613-0073
Attribution 4.0 International (CC BY 4.0).
CEUR Workshop Proceedings (CEUR-WS.org) altitude affects UAV performance in different weather
CEUR
ceur-ws.org
Workshop ISSN 1613-0073
Proceedings
conditions. For example, flying above or below certain intrusions. The system is integrated with exist-
weather layers (like fog or clouds) can be crucial. ing security infrastructure, providing a bird’s-eye
The optimized altitude can ensure communication view when a ground sensor is triggered.
with both the underwater sensor network and the control • General Atomics Predator B - used for na-
station [1]. tional border surveillance, can be used in con-
junction with ground sensor arrays for detecting
1.2. Speed optimization and tracking movements and is equipped with
high-resolution cameras and advanced signal in-
Generally, faster speeds increase energy consumption. telligence equipment that can integrate with sen-
The optimization algorithm should identify the most sor network data.
energy-efficient cruising speed for each UAV model.
Faster speeds allow for quicker coverage of an area but All the mentioned UAVs have a custom design naviga-
might reduce the effectiveness of sensors due to motion tion systems with included energy-efficient software algo-
blur or reduced processing time. rithms for routing and altitude/speed optimization, using
Speed must be optimized to balance routine surveil- various algorithms such as RL (Reinforcement Learn-
lance with the need for rapid response in case of detected ing, Dynamic Programming, Dijkstra, GA (Genetic algo-
threats [2, 3]. Tailwind can be exploited to reduce energy rithms) in different combinations.
consumption, whereas flying into headwinds will require
more energy, affecting optimal speed decisions.
3. Proposed solution
2. Related works The current research is focused on the development and
implementation of altitude (elevation) and speed opti-
There are some existing solutions related to the UAVs mization algorithm in custom designed UAVs.
assisted underwater perimeter security sensor networks The proposed algorithm is based on PSO (Particle
as: Swarm Optimization) [4, 5, 6]. This is a computational
method that optimizes a problem by iteratively trying
• DJI Enterprise Drones - the solution is used for to improve a candidate solution with regard to a given
inspection and surveillance of commercial and measure of quality.
military complexes. The drone is equipped with It solves a problem by having a population of candi-
thermal imaging sensors, high-resolution cam- date solutions, here dubbed particles, and moving these
eras, and programmable flight paths and is pro- particles around in the search-space according to simple
grammed for routine patrols or dispatched upon mathematical formulae over the particle’s position and
alerts from ground and underwater sensors. velocity.
• AeroVironment Raven RQ-11B - the solution Each particle’s movement is influenced by its local
is used for battlefield reconnaissance and surveil- best known position but is also guided toward the best
lance. The UAV is equipped with GA (Genetic known positions in the search-space, which are updated
Algorithms), based trajectory optimization sys- as better positions are found by other particles. When ap-
tem and interfaces with ground and underwater plying PSO for altitude and speed optimization in UAVs
control systems and sensor networks. supporting underwater perimeter security sensor net-
• Elbit Systems Skylark I-LEX – this is electri- works, the goal is to determine the optimal flight paths,
cally propelled UAV equipped with MPC (Model altitudes, and speeds for the UAVs to maximize coverage,
Predictive Control) trajectory optimization sys- efficiency, and responsiveness while minimizing energy
tem, designed to collect data and interface with consumption.
ground and underwater sensors for a comprehen-
sive security net and is utilized by military and 3.1. Challenges in the speed and elevation
homeland security for national borders and sen-
sitive areas.
optimization
• Anduril Industries’ Lattice – this is a complete The following challenges related to the speed/elevation
system that integrates drones, ground and under- optimization problem were defined during the research:
water sensors, and AI-powered analysis to detect,
classify, and respond to threats. • High Dimensionality: The speed/elevation opti-
mization problem can be high-dimensional, es-
• Asylon DroneCore - automated drone deploy-
pecially when considering 3D space and time,
ment system that works with perimeter sensors
making it computationally intensive [7].
to conduct autonomous patrols and respond to
• Dynamic Constraints: UAVs must respond to dy- where 𝜔 is the inertia weight, 𝑐1 and 𝑐2 are cognitive
namic changes in the environment, which re- and social coefficients, respectively, 𝑟1 , 𝑟2 are random
quires the PSO to be adaptable and responsive numbers between 0 and 1.
in real-time.
• Local Minima: The PSO algorithm may get Position update:
trapped in local minima. This issue can be miti- (𝑡)
gated by tuning the parameters (𝜔, 𝑐1 , 𝑐2 ) or by 𝑝𝑡+1
𝑖 = 𝑝𝑖 + 𝑣𝑖𝑡+1 . (2)
hybridizing PSO with other optimization tech- Ensure that the updated position adheres to the con-
niques. straints.
• Safety and Collision Avoidance: Ensuring safety
is paramount. The algorithm must incorporate Evaluation:
collision avoidance with other UAVs, terrain, and
obstacles [8]. Evaluate the fitness of each particle using the objective
function 𝑓 (𝑥).
Update the personal best pbest,i if the current position
3.2. Implementation of the particle yields a better value of the objective func-
mplementing a Particle Swarm Optimization (PSO) al- tion. Update the global best 𝑝𝑔𝑙𝑜𝑏𝑎𝑙 𝑏𝑒𝑠𝑡 if any particle
gorithm for altitude and speed optimization in UAV- achieves a better value than the current global best.
supported underwater perimeter security sensor net-
works involves several mathematical concepts. Here’s Termination:
an mathematical overview of the proposed algorithm
[9, 10, 11] : Continue iterating until a maximum number of iterations
is reached or convergence criteria are met (e.g., minimal
improvement in the global best).
Objective Function
Let’s denote the objective function as 𝑓 (𝑥), where 𝑥 Example Objective Function
represents a vector of the decision variables (altitude
and speed in this case) for UAVs. The function might aim Consider a simplified example where the objective is to
to minimize energy consumption while maximizing area minimize energy consumption 𝐸 while ensuring good
coverage, response time, or signal quality. area coverage 𝐶. The objective function might look like
This could be a weighted sum or a more complex func- this:
tion based on the mission requirements. 𝑓 (𝑥) = 𝛼𝐸 (𝑥) − 𝛽𝐶 (𝑥) . (3)
Here, 𝛼 and 𝛽 are weights reflecting the importance of
Constraints energy consumption versus coverage.
Include constraints like battery life (B), maximum and The functions 𝐸(𝑥) and 𝐶(𝑥) compute the energy
minimum altitude (𝐴{max} , 𝐴{min} ), and speed limits consumption and coverage based on the altitude and
(𝑆{max} , 𝑆{min} ). speed parameters in 𝑥.
The mathematical overview provided here is a sim-
plified version of what could be a complex real-world
PSO Algorithm Structure implementation. In practice, the functions and parame-
Particle Representation - each particle 𝑖 in the swarm rep- ters would need to be tailored to specific UAV capabilities,
resents a potential solution, with its position pi indicating sensor characteristics, environmental factors, and mis-
a particular set of altitudes and speeds for a UAV. sion goals.
Initialization: randomly initialize the position pi and Additionally, various enhancements to the basic PSO,
velocity vi of each particle within the feasible space de- such as constriction factors or varying inertia weight,
fined by the constraints. might be employed to improve convergence and solution
quality.
To implement the PSO algorithm for altitude and speed
Velocity and Position Update Rules
optimization in UAV-supported underwater perimeter
Velocity update: security sensor networks, we will develop a structured
(𝑡)
(︁
(𝑡)
)︁ pseudocode.
𝑣𝑖𝑡+1 = 𝜔𝑣𝑖 + 𝑐1 𝑟1 𝑝𝑏𝑒𝑠𝑡,𝑖 − 𝑝𝑖 This pseudocode will help visualization the flow of the
(︁ )︁ (1) algorithm and serve as a guide for actual programming.
(𝑡)
+ 𝑐2 𝑟2 𝑝𝑔𝑙𝑜𝑏𝑎𝑙 𝑏𝑒𝑠𝑡 − 𝑝𝑖 ,
Remember that PSO is inherently iterative and works Related to the pseudocode above please note:
with a population of solutions, adjusting them over time Initialization: The initial positions and velocities are
based on a defined objective function. randomly assigned within the permissible ranges for al-
The related PSO algorithm written in pseudocode is titude and speed. Each particle’s initial position is con-
shown below: sidered its personal best (pbest).
Updating Velocities and Positions: The velocities
PSO algorithm for UAVs elevation/speed are updated considering both the particle’s own best po-
optimization sition and the global best (gbest). The updated velocity
Inputs: influences the new position. It’s important to ensure that
- num_particles: Number of particles in the the updated positions are within the allowed ranges.
swarm Evaluating and Updating Best Positions: Af-
- max_iterations: Maximum number of ter updating positions, evaluate them using the objec-
iterations tive_function. If a particle’s new position is better than
- objective_function: Function to optimize its pbest, update pbest. If it’s better than the current gbest,
(minimize or maximize) update gbest.
- A_max, A_min: Maximum and minimum Termination: The algorithm iterates through this
allowable altitudes process, gradually moving the swarm towards the best
- S_max, S_min: Maximum and minimum solution. The process repeats either until the maximum
allowable speeds number of iterations is reached or some other stopping
- omega: Inertia weight criterion (like a convergence threshold) is met.
- c1, c2: Cognitive and social coefficients Returning the Optimal Solution: Finally, the gbest
Initialize: after the last iteration is returned as the optimal set of
- Create num_particles particles with random altitude and speed parameters.
positions and velocities Customization for the specific use-case: The ob-
- for each particle i: jective function should be designed specifically for the
- position[i] = Random within UAV’s operational requirements, taking into account fac-
[A_min, A_max] and [S_min, S_max] tors like energy consumption, area coverage, sensor ef-
- velocity[i] = Random initial fectiveness, and other mission-specific metrics.
velocity Parameters such as 𝜔, 𝑐1 , and 𝑐2 may need tuning for
- pbest[i] = position[i] optimal performance in specific scenarios.
- gbest = position of the best particle based Additional constraints or enhancements can be inte-
on objective_function grated into the algorithm based on specific requirements
Main Loop: and operational environments.
- for iter = 1 to max_iterations:
- for each particle i:
- Update velocity: 4. Key Takeaways
- r1, r2 = Random numbers
between 0 and 1 Enhanced Efficiency: The PSO algorithm effectively
- velocity[i] = omega * velocity[i] optimizes UAV flight parameters (altitude and speed),
+ c1 * r1 * (pbest[i] - position[i]) leading to improved energy efficiency. This results in
+ c2 * r2 * (gbest - position[i]) longer mission durations and reduced operational costs.
- Update position: Adaptive Flight Paths: The algorithm’s ability to
- position[i] = position[i] dynamically adapt flight paths in response to changing
+ velocity[i] environmental conditions and mission requirements is
- Ensure position[i] adheres to a significant advantage, ensuring optimal coverage and
[A_min, A_max] and [S_min, S_max] data collection.
- Evaluate: Collaborative Functionality: PSO inherently sup-
- If objective_function(position[i]) is ports multi-UAV coordination, allowing for effective
swarm operations. This results in comprehensive area
better than objective_function(pbest[i]):
- pbest[i] = position[i] surveillance and redundant systems for critical defense
- If objective_function(position[i]) is missions.
better than objective_function(gbest): Real-Time Decision Making: The implementation
- gbest = position[i] enables UAVs to make real-time adjustments, crucial for
- Return gbest as the optimal solution responding to emergent underwater threats or anomalies
End Algorithm detected by the sensor network.
Operational Flexibility: The algorithm’s flexibil- and maritime technologies, paving the way for enhanced
ity allows it to be tailored to various mission scenarios, security solutions in coastal and offshore environments.
UAV types, and sensor network configurations, making In conclusion, PSO is a robust and versatile algorithm
it broadly applicable in underwater perimeter defense. widely used for solving complex optimization problems.
Its ongoing developments and applications across diverse
4.1. Challenges and Considerations fields highlight its relevance in the current technological
landscape.
Complex Environmental Dynamics: The underwa-
ter and aerial environments present unique challenges,
including variable weather conditions and underwater References
currents, which can affect the algorithm’s performance.
Communication Limitations: Ensuring reliable [1] H. He, S. Zhang, Y. Zeng, R. Zhang, Joint altitude
communication between UAVs and underwater sensors and beamwidth optimization for uav-enabled mul-
remains a challenge, impacting the coordination and ef- tiuser communications, IEEE Communications Let-
fectiveness of the network. ters 22 (2017) 344–347.
Computational Demands: PSO, especially in real- [2] N. H. Chu, D. T. Hoang, D. N. Nguyen,
time applications, can be computationally intensive, ne- N. Van Huynh, E. Dutkiewicz, Joint speed control
cessitating robust onboard processing capabilities. Secu- and energy replenishment optimization for uav-
rity and Robustness: The system must be secured against assisted iot data collection with deep reinforcement
potential cyber threats and robust enough to handle op- transfer learning, IEEE Internet of Things Journal
erational uncertainties and potential system failures. 10 (2022) 5778–5793.
[3] W. Ye, J. Luo, F. Shan, W. Wu, M. Yang, Offspeeding:
Optimal energy-efficient flight speed scheduling for
5. Conclusion uav-assisted edge computing, Computer Networks
183 (2020) 107577.
The implementation of a Particle Swarm Optimization [4] D. Wang, D. Tan, L. Liu, Particle swarm optimiza-
(PSO) algorithm for altitude and speed optimization in tion algorithm: an overview, Soft computing 22
UAVs supporting underwater perimeter security sensor (2018) 387–408.
networks is a sophisticated approach that leverages the [5] O. F. Aje, A. A. Josephat, The particle swarm op-
strengths of swarm intelligence for operational efficiency. timization (pso) algorithm application–a review,
The conclusion drawn from this implementation can Global Journal of Engineering and Technology Ad-
highlight its significance, potential benefits, and areas vances 3 (2020) 001–006.
for future enhancement. Future steps: [6] M. Ehteram, A. Seifi, F. B. Banadkooki, Structure of
particle swarm optimization (pso), in: Hellenic Con-
• incorporating advanced variants of PSO or hybrid
ference on Artificial Intelligence, Springer, 2022, pp.
algorithms could further optimize performance,
23–32.
especially in highly dynamic or unpredictable
[7] S. Schräder, D. Schleich, S. Behnke, Two-step plan-
environments.
ning of dynamic uav trajectories using iterative-
• leveraging AI for predictive analytics and ma- spaces, in: International Conference on Intelligent
chine learning for continuous improvement of Autonomous Systems, Springer, 2022, pp. 257–271.
flight path algorithms based on historical data [8] J. Hong, D. Chen, W. Li, Z. Fan, Trajectory plan-
can enhance operational efficiency. ner for uavs based on potential field obtained by a
• incorporating sustainable technologies, such as kinodynamic gene regulation network, Sensors 23
solar-powered UAVs, can extend mission dura- (2023) 7982.
tions and reduce environmental impact. [9] R. Poli, J. Kennedy, T. Blackwell, Particle swarm
intelligence. an overview, Swarm Intelligence 1
The implementation of a PSO algorithm for optimiz-
(2007) 33–57.
ing a group of UAVs’ altitude and speed in underwater
[10] Y. Yang, X. Xiong, Y. Yan, Uav formation trajectory
perimeter security sensor networks demonstrates sig-
planning algorithms: A review, Drones 7 (2023) 62.
nificant potential in improving maritime security oper-
[11] Q. Geng, Z. Zhao, A kind of route planning method
ations [12]. While challenges remain, the continuous
for uav based on improved pso algorithm, in:
advancements in technology and algorithmic strategies
2013 25th Chinese control and decision conference
hold promise for developing more sophisticated, efficient,
(CCDC), IEEE, 2013, pp. 2328–2331.
and robust defense networks in the future . This ap-
[12] H. Zhang, B. Xin, L.-h. Dou, J. Chen, K. Hirota, A re-
proach exemplifies the innovative integration of aerial
view of cooperative path planning of an unmanned
aerial vehicle group, Frontiers of Information Tech-
nology & Electronic Engineering 21 (2020) 1671–
1694.