Active security

Abstract

In this paper we introduce active security, a new methodology which introduces programmatic control within a novel feedback loop into the defense infrastructure. Active security implements a unified programming environment which provides interfaces to (i) protect the infrastructure under common attack scenarios (e.g., configure a firewall), (ii) sense the current state of the infrastructure through a wide variety of information, (iii) adjust the configuration of the infrastructure at run time based on sensed information, (iv) collect forensic evidence on-demand, at run-time for attribution, and (v) counter the attack through more advanced mechanisms such as migrating malicious code to a quarantined system. We built an initial prototype that extends the FloodLight software-defined networking controller to automatically interface with the Snort intrusion detection system to detect anomalies, the Linux Memory Extractor to collect forensic evidence at run-time, and the Volatility parsing tool to extract an executable from physical memory and analyze information about the malware (which can then be used by the active security system to better secure the infrastructure).