Detecting Stealthy Ransomware in IPFS Networks Using Machine Learning
J.J. Chen, Guirong Zhang
Abstract
Ransomware remains one of the most pernicious threats in cybersecurity, with its distribution mechanisms evolving alongside technological advancements. This study explores the efficacy of machine learning techniques in detecting ransomware activities within the InterPlanetary File System (IPFS), a decentralized storage network. The research evaluates several machine learning models, including Logistic Regression, Decision Trees, Random Forests, Gradient Boosting Machines, and Convolutional Neural Networks, to assess their accuracy, precision, recall, and robustness under adversarial conditions. Results indicate that advanced models, particularly Convolutional Neural Networks and Random Forests, perform with high effectiveness, maintaining substantial accuracy and resilience against evasion techniques. The findings underscore the potential of integrating machine learning into cybersecurity measures for decentralized systems, proposing a promising avenue for enhancing IPFS's resistance to ransomware threats. Future work should focus on expanding dataset diversity, improving model adaptability to new and evolving threats, and assessing the deployment feasibility in varied operational contexts.