Vertical Federated Learning Based Privacy-Preserving Cooperative Sensing in Cognitive Radio Networks

Research output: Chapter in Book/Report/Conference proceedingConference paperpeer-review

17 Citations (Scopus)

Abstract

Machine learning-based cooperative sensing scheme, despite its effectiveness in significantly improving the sensing performance, suffers from privacy threats because the sensing reports shared by secondary users (SUs) are highly correlated to their locations, which can be maliciously exploited to infer private information. In this paper, we propose a novel vertical federated learning-based cooperative sensing (VFL-CS) scheme where sensing results are kept locally at each smart SU (SSU) and the model is trained in a decentralised collaborative learning setting. A multi-user deep learning-based FL architecture is constructed with detailed training and evaluation processes explained and security analysed. Simulation results show that our proposed VFL-CS scheme outperforms conventional soft-fusion based cooperative sensing (SF-CS) scheme in terms of much higher area under curve (AUC) score with high data privacy-preserving capability.

Original languageEnglish
Title of host publication2020 IEEE Globecom Workshops, GC Wkshps 2020 - Proceedings
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781728173078
DOIs
Publication statusPublished - Dec 2020
Event2020 IEEE Globecom Workshops, GC Wkshps 2020 - Virtual, Taipei, Taiwan
Duration: 7 Dec 202011 Dec 2020

Publication series

Name2020 IEEE Globecom Workshops, GC Wkshps 2020 - Proceedings

Conference

Conference2020 IEEE Globecom Workshops, GC Wkshps 2020
Country/TerritoryTaiwan
CityVirtual, Taipei
Period7/12/202011/12/2020

Keywords

  • additively homomorphic encryption
  • Cognitive radio
  • deep learning
  • federated learning
  • secure cooperative sensing

Fingerprint

Dive into the research topics of 'Vertical Federated Learning Based Privacy-Preserving Cooperative Sensing in Cognitive Radio Networks'. Together they form a unique fingerprint.

Cite this