English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

In the last years, the number of IoT devices deployed has suffered anundoubted explosion, reaching the scale of billions. However, some newcybersecurity issues have appeared together with this development. Some ofthese issues are the deployment of unauthorized devices, malicious codemodification, malware deployment, or vulnerability exploitation. This fact hasmotivated the requirement for new device identification mechanisms based onbehavior monitoring. Besides, these solutions have recently leveraged Machineand Deep Learning techniques due to the advances in this field and the increasein processing capabilities. In contrast, attackers do not stay stalled and havedeveloped adversarial attacks focused on context modification and ML/DLevaluation evasion applied to IoT device identification solutions. This workexplores the performance of hardware behavior-based individual deviceidentification, how it is affected by possible context- and ML/DL-focusedattacks, and how its resilience can be improved using defense techniques. Inthis sense, it proposes an LSTM-CNN architecture based on hardware performancebehavior for individual device identification. Then, previous techniques havebeen compared with the proposed architecture using a hardware performancedataset collected from 45 Raspberry Pi devices running identical software. TheLSTM-CNN improves previous solutions achieving a +0.96 average F1-Score and 0.8minimum TPR for all devices. Afterward, context- and ML/DL-focused adversarialattacks were applied against the previous model to test its robustness. Atemperature-based context attack was not able to disrupt the identification.However, some ML/DL state-of-the-art evasion attacks were successful. Finally,adversarial training and model distillation defense techniques are selected toimprove the model resilience to evasion attacks, without degrading itsperformance.

Adversarial attacks and defenses on ML- and hardware-based IoT device fingerprinting and identification