Detection of Malicious Network Traffic Behavior Using JA3 Fingerprints

Abstract

This paper presents a novel approach for classifying spoof network traffic based on JA3 fingerprint clustering. In particular, it concerns the detection of so-called zero-day malware. The proposed method does not work with known JA3 hashes. However, it compares the JA3 fingerprint of captured traffic with JA3 fingerprints of traffic with predefined criteria, such as the use of current cipher suites or protocol, for classification.