VPNDroid: Malicious Android VPN detection using a CNN-RF method

Nikolaos Polatidis, Elias Pimenidis, Marcello Trovati, Lazaros Iliadis

Research output: Chapter in Book/Conference proceeding with ISSN or ISBNConference contribution with ISSN or ISBNpeer-review

Abstract

Protecting online privacy using Virtual Private Networks (VPNs) is not as simple as it seems, since many well-known VPNs may not be secure. Despite appearing to be secure on the surface, VPNs can be a complete privacy and security disaster by stealing bandwidth, infecting devices with malware, installing tracking libraries, stealing personal data, and leaving data exposed to third parties. Therefore, Android users must exercise caution when downloading and installing VPN software on their devices. To this end, this paper proposes a neural network combined with a random forest that identifies malicious and malware-infected VPNs based on app permissions, along with a novel dataset of malicious and benign Android VPNs. The experimental results demonstrate that our classifier achieves high accuracy and outperforms other standard classifiers in terms of evaluation metrics such as accuracy, precision, and recall.
Original languageEnglish
Title of host publicationArtificial Neural Networks and Machine Learning
Subtitle of host publicationICANN 2023
Pages444–453
ISBN (Electronic)9783031442049
DOIs
Publication statusPublished - 22 Sept 2023
Event32nd International Conference on Artificial Neural Networks - Astoria Capsis Hotel, Crete, Greece
Duration: 26 Sept 202329 Sept 2023
Conference number: 32
https://e-nns.org/icann2023/

Publication series

NameLecture Notes in Computer Science
Volume14263

Conference

Conference32nd International Conference on Artificial Neural Networks
Abbreviated titleICANN 2023
Country/TerritoryGreece
CityCrete
Period26/09/2329/09/23
Internet address

Fingerprint

Dive into the research topics of 'VPNDroid: Malicious Android VPN detection using a CNN-RF method'. Together they form a unique fingerprint.

Cite this