Ultrafast Cognitive Edge‐Empowered UAV Security System With Federated Learning for Energy‐Efficient Antieavesdropping in High‐Mobility Networks.
Saved in:
| Title: | Ultrafast Cognitive Edge‐Empowered UAV Security System With Federated Learning for Energy‐Efficient Antieavesdropping in High‐Mobility Networks. |
|---|---|
| Authors: | Kadhim, Mohammed Aboud1 (AUTHOR) mohammed_aboud@mtu.edu.iq, Hamid, Hamood Shehab2 (AUTHOR), Aftan, Ahmed Obaid2 (AUTHOR), Shao, Zijian (AUTHOR) sbinwal@wiley.com |
| Source: | International Journal of Antennas & Propagation. 4/5/20226, Vol. 2026, p1-19. 19p. |
| Subjects: | Federated learning, Eavesdropping, Machine learning, Edge computing, Drone warfare, Wireless communications, Energy consumption |
| Abstract: | The unmanned aerial vehicle (UAV) network is one of the most crucial types of wireless networks that encounter security problems in highly dynamic environments, and in such cases, centralized approaches cannot ensure real‐time threat detection and robust energy‐efficient antieavesdropping capabilities. In response, this letter proposes a cognitive edge‐empowered ultrafast UAV security system, utilizing federated learning for highly dynamic adaptive antieavesdropping communication over UAV networks. We propose a framework that leverages several novel components including (1) 50ms ultrashort time slots yielding 20 × faster response than conventional 1s‐based approaches, (2) continuous online federated learning with subsecond model updates over distributed UAV nodes, (3) predictive mobility modeling to proactively make security decisions for UAVs that can fly at speeds of up to 50m/s, and (4) adaptive time scaling, where model parameters can be adjusted online based on the mobility pattern and threat level. The system consists of source UAVs (SUAVs) that convey communication services and jamming UAVs (JUAVs) that deploy proactive countermeasures while leveraging usage of a 22‐dimensional feature space for ML‐based threat detection. In a UAV environment, where data under uncertain and non‐IID conditions are prevalent, these conditions have led to a robust model aggregation by the confidence‐weighted federated averaging algorithm. Extensive simulations validate our design with results including 88.3% prediction accuracy in eavesdropper behavior detection, average response time 85ms (98% of time realizable in real time under 100ms delay threshold), and 2.67 bits/J secrecy energy efficiency (a 75% improvement over the traditional methods). It successfully preserves performance relatively high on 5‐50 UAV networks with slight performance loss, while adaptive A federated learning framework achieves fast convergence through momentum‐based optimization and real‐time model federalization. [ABSTRACT FROM AUTHOR] |
| Copyright of International Journal of Antennas & Propagation is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Database: | Engineering Source |
|
Full text is not displayed to guests.
Login for full access.
|
|
| Abstract: | The unmanned aerial vehicle (UAV) network is one of the most crucial types of wireless networks that encounter security problems in highly dynamic environments, and in such cases, centralized approaches cannot ensure real‐time threat detection and robust energy‐efficient antieavesdropping capabilities. In response, this letter proposes a cognitive edge‐empowered ultrafast UAV security system, utilizing federated learning for highly dynamic adaptive antieavesdropping communication over UAV networks. We propose a framework that leverages several novel components including (1) 50ms ultrashort time slots yielding 20 × faster response than conventional 1s‐based approaches, (2) continuous online federated learning with subsecond model updates over distributed UAV nodes, (3) predictive mobility modeling to proactively make security decisions for UAVs that can fly at speeds of up to 50m/s, and (4) adaptive time scaling, where model parameters can be adjusted online based on the mobility pattern and threat level. The system consists of source UAVs (SUAVs) that convey communication services and jamming UAVs (JUAVs) that deploy proactive countermeasures while leveraging usage of a 22‐dimensional feature space for ML‐based threat detection. In a UAV environment, where data under uncertain and non‐IID conditions are prevalent, these conditions have led to a robust model aggregation by the confidence‐weighted federated averaging algorithm. Extensive simulations validate our design with results including 88.3% prediction accuracy in eavesdropper behavior detection, average response time 85ms (98% of time realizable in real time under 100ms delay threshold), and 2.67 bits/J secrecy energy efficiency (a 75% improvement over the traditional methods). It successfully preserves performance relatively high on 5‐50 UAV networks with slight performance loss, while adaptive A federated learning framework achieves fast convergence through momentum‐based optimization and real‐time model federalization. [ABSTRACT FROM AUTHOR] |
|---|---|
| ISSN: | 16875869 |
| DOI: | 10.1155/ijap/5533666 |
