Automated Detection of Drive-By Download Attack

To extract features of Drive-by-download traffic, we develop a system for visualizing the flow of HTTP traffic. Based on the typical behavior observed in malicious connections, we propose an automated detection system to classify drive-by download with the logical condition of some features such as the redirection methods and the object size. Our detection uses the decision tree learning with optimal threshold for learning data and successfully detect the malicious connection with the accuracy of 0.06 false negative and 0 false positive.