PulseExploreJournal ClubDebatesTrendingResearchersJournals
Instagram
HomeExploreJournal ClubTrending
Synapse
⌘+K
Synapse
May 22, 2026Applied SciencesOpen Access

Application-Oriented Evaluation of Federated Learning for IoT Intrusion Detection Under Non-IID Conditions in Wireless Sensor Networks

View Full Paper
Ask AI
Bookmark
Share

Authors

WAWalaa AlayedHTHassan TahirWHWaqar Ul Hassan

Discussion

Loading...

Member takes

Overview

Randomized trial evaluates intrusion detection efficiency in IoT, suggesting effective aggregation strategies.

Key Points

  • This research aims to evaluate the effectiveness of federated learning in intrusion detection under non-IID conditions.
  • Trained an LSTM-based model for intrusion detection in a federated setting.
  • Assessed using datasets WSN-DS, CIC-IDS-2017, and UNSW-NB15.
  • Evaluated aggregation strategies: FedAvg, FedProx, and SCAFFOLD under varying skew scenarios.
  • FedAvg showed accuracy reductions of up to 23.4 percentage points in extreme non-IID conditions.
  • FedProx and SCAFFOLD improved convergence stability and reduced client drift impact.
  • SCAFFOLD achieved up to 45% lower communication cost compared to FedAvg.

Cite This Study

Alayed et al. (2026) studied this question.

synapsesocial.com/papers/6a0ff327d674f7c03778bab4https://doi.org/10.3390/app16105092
View Full Paper
Ask AI
Bookmark
Share