Fighting Under-price DoS Attack in Ethereum with Machine Learning Techniques
José Eduardo de Azevedo Sousa, Vinícius Cunha Oliveira, Júlia Almeida Valadares, Alex Borges Vieira, Heder S. Bernardino, Saulo Moraes Villela, Glauber Dias Gonçalves · ACM SIGMETRICS Performance Evaluation Review · 2021
Ethereum is one of the most popular cryptocurrency currently and it has been facing security threats and attacks. As a consequence, Ethereum users may experience long periods to validate transactions. Despite the maintenance on the Ethereum mechanisms, there are still indications that it remains susceptible to a sort of attacks. In this work, we analyze the Ethereum network behavior during an under-priced DoS attack, where malicious users try to perform denial-of-service attacks that exploit flaws in the fee mechanism of this cryptocurrency. We propose the application of machine learning techniques and ensemble methods to detect this attack, using the available transaction attributes. The proposals present notable performance as the Decision Tree models, with AUC-ROC, F-score and recall larger than 0.94, 0.82, and 0.98, respectively.