Towards Automated Threat Intelligence Fusion

Abstract

The volume and frequency of new cyber attacks have exploded in recent years. Such events have very complicated workflows and involve multiple criminal actors and organizations. However, current practices for threat analysis and intelligence discovery are still performed piecemeal in an ad-hoc manner. For example, a modern malware analysis system can dissect a piece of malicious code by itself. But, it cannot automatically identify the criminals who developed it or relate other cyber attack events with it. Consequently, it is imperative to automatically assemble the jigsaw puzzles of cybercrime events by performing threat intelligence fusion on data collected from heterogeneous sources, such as malware, underground social networks, cryptocurrency transaction records, etc. In this paper, we propose an Automated Threat Intelligence fuSion framework (ATIS) that is able to take all sorts of threat sources into account and discover new intelligence by connecting the dots of apparently isolated cyber events. To this end, ATIS consists of 5 planes, namely analysis, collection, controller, data and application planes. We discuss the design choices we made in the function of each plane and the interfaces between two adjacent planes. In addition, we develop two applications on top of ATIS to demonstrate its effectiveness.