Wide spectrum attribution: Using deception for attribution intelligence in cyber attacks
Modern cyber attacks have evolved considerably. The skill level required to conduct a cyber attack is low. Computing power is cheap, targets are diverse and plentiful. Point-and-click crimeware kits are widely circulated in the underground economy, while source code for sophisticated malware such as Stuxnet is available for all to download and repurpose. Despite decades of research into defensive techniques, such as firewalls, intrusion detection systems, anti-virus, code auditing, etc, the quantity of successful cyber attacks continues to increase, as does the number of vulnerabilities identified. Measures to identify perpetrators, known as attribution, have existed for as long as there have been cyber attacks. The most actively researched technical attribution techniques involve the marking and logging of network packets. These techniques are performed by network devices along the packet journey, which most often requires modification of existing router hardware and/or software, or the inclusion of additional devices. These modifications require wide-scale infrastructure changes that are not only complex and costly, but invoke legal, ethical and governance issues. The usefulness of these techniques is also often questioned, as attack actors use multiple stepping stones, often innocent systems that have been compromised, to mask the true source. As such, this thesis identifies that no publicly known previous work has been deployed on a wide-scale basis in the Internet infrastructure. This research investigates the use of an often overlooked tool for attribution: cyber de- ception. The main contribution of this work is a significant advancement in the field of deception and honeypots as technical attribution techniques. Specifically, the design and implementation of two novel honeypot approaches; i) Deception Inside Credential Engine (DICE), that uses policy and honeytokens to identify adversaries returning from different origins and ii) Adaptive Honeynet Framework (AHFW), an introspection and adaptive honeynet framework that uses actor-dependent triggers to modify the honeynet envi- ronment, to engage the adversary, increasing the quantity and diversity of interactions. The two approaches are based on a systematic review of the technical attribution litera- ture that was used to derive a set of requirements for honeypots as technical attribution techniques. Both approaches lead the way for further research in this field.
- PhD