EA-POT: An Explainable AI Assisted Blockchain Framework for Honeypot IP Predictions

Abstract

The culpable cybersecurity practices that threaten leading organizations are logically prone to establishing countermeasures, including honeypots, and bestowing research innovations in various dimensions, such as ML-enabled threat predictions. This article proposes an explainable AI-assisted permissioned blockchain framework named EA-POT for predicting potential defaulters' IP addresses. EA-POT registers the probable defaulters predicted by explainable AI based on the approval of IP authorizers of blockchain databases. Experiments were carried out at the IoT Cloud Research laboratory using three prediction models, such as Random Forest Modeling (RFM), Linear Regression Modeling (LRM), and Support Vector Machines (SVM); and, the experimental results for predicting the AWS honeypots were explored. The proposed EA-POT framework revealed the procedure for including interpretable knowledge while blacklisting IPs that reach honeypots.