Mitigating Phishing Threats in Unmanned Aircraft Systems (UAS) Through Multi-Stage Defense Strategies

Abstract

Phishing attacks, while more commonly associated with targeting individuals or organizations through traditional communication channels like email or social media, pose potential threats to unmanned aircraft systems (UAS) or drones. Although not as prevalent in this domain, there exist scenarios where phishing tactics could compromise UAS operations and data. Attackers might impersonate legitimate UAS entities, crafting emails that appear credible and relevant to UAS operations. A multi-stage approach incorporating natural language processing and machine learning is introduced to combat such threats. This approach employs techniques like conditional random field (CRF) and latent Dirichlet allocation (LDA) to detect phishing attacks and discern manipulated content. A novel web crawler utilizing web ontology language (OWL) is devised, leveraging semantic relationships to filter out fake sites from search results. The experimental results demonstrate the effectiveness of these methods in detecting and preventing phishing attacks across different platforms.