Cybersecurity“Instant replay” quickly pinpoints cyberattack details

Until now, assessing the extent and impact of network or computer system attacks has been largely a time-consuming manual process. A new software system being developed by cybersecurity researchers at the Georgia Institute of Technology will largely automate that process, allowing investigators to quickly and accurately pinpoint how intruders entered the network, what data they took and which computer systems were compromised.

Georgia Tech says that the system, known as Refinable Attack INvestigation (RAIN), will provide forensic investigators a detailed record of an intrusion, even if the attackers attempted to cover their tracks. The system provides multiple levels of detail, facilitating automated searches through information at a high level to identify the specific events for which more detailed data is reproduced and analyzed.

“You can go back and find out what has gone wrong in your system, not just at the point where you realized that something is wrong, but far enough back to figure out how the attacker got into the system and what has been done,” said Wenke Lee, co-director of Georgia Tech’s Institute for Information Security & Privacy. 

The research, supported largely by the Defense Advanced Research Projects Agency (DARPA) and also by the National Science Foundation and Office of Naval Research, is scheduled to be reported 31 October at the 2017 ACM Conference on Computer and Communications Security (CCS).

Existing forensic techniques can provide detailed information about the current status of computers and networks; from that information, investigators can then attempt to infer how attacks unfolded. Digital logs maintained by the systems provide some information about attacks, but because of concerns about data storage issues, usually don’t record enough detail. Other programs provide snapshots in time, but those snapshots may miss important details of an attack.

The RAIN system continuously monitors a system and logs events that it recognizes as potentially interesting. That ability to selectively record information likely to be useful later allows a trade-off between realistic overhead – in terms of system performance and data storage – and useful levels of detail. The system “effectively prunes out unrelated processes and determines attack causality with negligible false positive rates,” the authors wrote in their conference paper.