Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Distributed Intrusion Detection in Dynamic Networks of UAVs using Few-Shot Federated Learning

Ozlem Ceviz, Sevil Sen, Pinar Sadioglu

TL;DR

This work tackles intrusion detection in dynamic FANETs where privacy, data scarcity, and lossy communications hinder traditional centralized and large-data approaches. It proposes FSFL-IDS, a hybrid framework that merges Federated Learning with Few-shot Learning, leveraging FedAvg for distributed model aggregation while adapting quickly to scarce labeled data via N-way K-shot strategies and Hyperband-driven hyperparameter optimization. Experimental evaluation on a FANET-like Ns-3 setup shows that FSFL-IDS achieves competitive accuracy to FL-IDS while using only about 10% of the data and reducing communication rounds, thereby lowering energy consumption and latency in challenging UAV environments. The results demonstrate FSFL-IDS’ practicality for resource-constrained, highly dynamic networks and support its potential extension to other mobile ad hoc and IoT scenarios, while acknowledging limitations such as data heterogeneity and potential adversarial updates.

Abstract

Flying Ad Hoc Networks (FANETs), which primarily interconnect Unmanned Aerial Vehicles (UAVs), present distinctive security challenges due to their distributed and dynamic characteristics, necessitating tailored security solutions. Intrusion detection in FANETs is particularly challenging due to communication costs, and privacy concerns. While Federated Learning (FL) holds promise for intrusion detection in FANETs with its cooperative and decentralized model training, it also faces drawbacks such as large data requirements, power consumption, and time constraints. Moreover, the high speeds of nodes in dynamic networks like FANETs may disrupt communication among Intrusion Detection Systems (IDS). In response, our study explores the use of few-shot learning (FSL) to effectively reduce the data required for intrusion detection in FANETs. The proposed approach called Few-shot Federated Learning-based IDS (FSFL-IDS) merges FL and FSL to tackle intrusion detection challenges such as privacy, power constraints, communication costs, and lossy links, demonstrating its effectiveness in identifying routing attacks in dynamic FANETs.This approach reduces both the local models and the global model's training time and sample size, offering insights into reduced computation and communication costs and extended battery life. Furthermore, by employing FSL, which requires less data for training, IDS could be less affected by lossy links in FANETs.

Distributed Intrusion Detection in Dynamic Networks of UAVs using Few-Shot Federated Learning

TL;DR

This work tackles intrusion detection in dynamic FANETs where privacy, data scarcity, and lossy communications hinder traditional centralized and large-data approaches. It proposes FSFL-IDS, a hybrid framework that merges Federated Learning with Few-shot Learning, leveraging FedAvg for distributed model aggregation while adapting quickly to scarce labeled data via N-way K-shot strategies and Hyperband-driven hyperparameter optimization. Experimental evaluation on a FANET-like Ns-3 setup shows that FSFL-IDS achieves competitive accuracy to FL-IDS while using only about 10% of the data and reducing communication rounds, thereby lowering energy consumption and latency in challenging UAV environments. The results demonstrate FSFL-IDS’ practicality for resource-constrained, highly dynamic networks and support its potential extension to other mobile ad hoc and IoT scenarios, while acknowledging limitations such as data heterogeneity and potential adversarial updates.

Abstract

Flying Ad Hoc Networks (FANETs), which primarily interconnect Unmanned Aerial Vehicles (UAVs), present distinctive security challenges due to their distributed and dynamic characteristics, necessitating tailored security solutions. Intrusion detection in FANETs is particularly challenging due to communication costs, and privacy concerns. While Federated Learning (FL) holds promise for intrusion detection in FANETs with its cooperative and decentralized model training, it also faces drawbacks such as large data requirements, power consumption, and time constraints. Moreover, the high speeds of nodes in dynamic networks like FANETs may disrupt communication among Intrusion Detection Systems (IDS). In response, our study explores the use of few-shot learning (FSL) to effectively reduce the data required for intrusion detection in FANETs. The proposed approach called Few-shot Federated Learning-based IDS (FSFL-IDS) merges FL and FSL to tackle intrusion detection challenges such as privacy, power constraints, communication costs, and lossy links, demonstrating its effectiveness in identifying routing attacks in dynamic FANETs.This approach reduces both the local models and the global model's training time and sample size, offering insights into reduced computation and communication costs and extended battery life. Furthermore, by employing FSL, which requires less data for training, IDS could be less affected by lossy links in FANETs.
Paper Structure (18 sections, 3 equations, 4 figures, 6 tables)

This paper contains 18 sections, 3 equations, 4 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Federated Learning-based IDS for FANETs
  4. Few-shot Learning-based IDS
  5. BACKGROUND
  6. Federated Learning
  7. Few-shot Learning
  8. Hyperband Algorithm
  9. THE PROPOSED APPROACH
  10. Few-shot Federated Learning-based IDS
  11. Experimental Settings
  12. Experimental Results
  13. Classifier: DNN vs CNN
  14. Effectiveness: L-IDS, FSFL-IDS and FL-IDS
  15. Efficiency: FSFL-IDS vs FL-IDS
  16. ...and 3 more sections

Figures (4)

  • Figure 1: Few-Shot Federated Learning-based IDS (FSFL-IDS)
  • Figure 2: Comparison of FL-IDS and FSFL-IDS in Detecting Sinkhole Attack
  • Figure 3: Comparison of FL-IDS and FSFL-IDS in Detecting Blackhole Attack
  • Figure 4: Comparison of FL-IDS and FSFL-IDS in Detecting Flooding Attack