Biblio

Using stolen or weak credentials to bypass authentication is one of the top 10 network threats, as shown in recent studies. Disguising as legitimate users, attackers use stealthy techniques such as rootkits and covert channels to gain persistent access to a target system. However, such attacks are often detected after the system misuse stage, i.e., the attackers have already executed attack payloads such as: i) stealing secrets, ii) tampering with system services, and ii) disrupting the availability of production services.

In this talk, we analyze a real-world credential stealing attack observed at the National Center for Supercomputing Applications. We show the disadvantages of traditional detection techniques such as signature-based and anomaly-based detection for such attacks. Our approach is a complement to existing detection techniques. We investigate the use of Probabilistic Graphical Model, specifically Factor Graphs, to integrate security logs from multiple sources for a more accurate detection. Finally, we propose a security testbed architecture to: i) simulate variants of known attacks that may happen in the future, ii) replay such attack variants in an isolated environment, and iii) collect and share security logs of such replays for the security research community.

Recent attacks show that threats to cyber infrastructure are not only increasing in volume, but are getting more sophisticated. The attacks may comprise multiple actions that are hard to differentiate from benign activity, and therefore common detection techniques have to deal with high false positive rates. Because of the imperfect performance of automated detection techniques, responses to such attacks are highly dependent on human-driven decision-making processes. While game theory has been applied to many problems that require rational decisionmaking, we find limitation on applying such method on security games. In this work, we propose Q-Learning to react automatically to the adversarial behavior of a suspicious user to secure the system. This work compares variations of Q-Learning with a traditional stochastic game. Simulation results show the possibility of Naive Q-Learning, despite restricted information on opponents.

This paper presents a framework for (1) generating variants of known attacks, (2) replaying attack variants in an isolated environment and, (3) validating detection capabilities of attack detection techniques against the variants. Our framework facilitates reproducible security experiments. We generated 648 variants of three real-world attacks (observed at the National Center for Supercomputing Applications at the University of Illinois). Our experiment showed the value of generating attack variants by quantifying the detection capabilities of three detection methods: a signature-based detection technique, an anomaly-based detection technique, and a probabilistic graphical model-based technique.

Today's cyber-physical systems (CPSs) can have very different characteristics in terms of control algorithms, configurations, underlying infrastructure, communication protocols, and real-time requirements. Despite these variations, they all face the threat of malicious attacks that exploit the vulnerabilities in the cyber domain as footholds to introduce safety violations in the physical processes. In this paper, we focus on a class of attacks that impact the physical processes without introducing anomalies in the cyber domain. We present the common challenges in detecting this type of attacks in the contexts of two very different CPSs (i.e., power grids and surgical robots). In addition, we present a general principle for detecting such cyber-physical attacks, which combine the knowledge of both cyber and physical domains to estimate the adverse consequences of malicious activities in a timely manner.

Many current VM monitoring approaches require guest OS modifications and are also unable to perform application level monitoring, reducing their value in a cloud setting. This paper introduces hprobes, a framework that allows one to dynamically monitor applications and operating systems inside a VM. The hprobe framework does not require any changes to the guest OS, which avoids the tight coupling of monitoring with its target. Furthermore, the monitors can be customized and enabled/disabled while the VM is running. To demonstrate the usefulness of this framework, we present three sample detectors: an emergency detector for a security vulnerability, an application watchdog, and an infinite-loop detector. We test our detectors on real applications and demonstrate that those detectors achieve an acceptable level of performance overhead with a high degree of flexibility.

This paper presents a Factor Graph based framework called AttackTagger for highly accurate and preemptive detection of attacks, i.e., before the system misuse. We use secu- rity logs on real incidents that occurred over a six-year pe- riod at the National Center for Supercomputing Applica- tions (NCSA) to evaluate AttackTagger. Our data consist of security incidents that led to compromise of the target system, i.e., the attacks in the incidents were only identified after the fact by security analysts. AttackTagger detected 74 percent of attacks, and the majority them were detected before the system misuse. Finally, AttackTagger uncovered six hidden attacks that were not detected by intrusion de- tection systems during the incidents or by security analysts in post-incident forensic analysis.

This talk will explore a scalable data analytics pipeline for real-time attack detection through the use of customized honeypots at the National Center for Supercomputing Applications (NCSA). Attack detection tools are common and are constantly improving, but validating these tools is challenging. You must: (i) identify data (e.g., system-level events) that is essential for detecting attacks, (ii) extract this data from multiple data logs collected by runtime monitors, and (iii) present the data to the attack detection tools. On top of this, such an approach must scale with an ever-increasing amount of data, while allowing integration of new monitors and attack detection tools. All of these require an infrastructure to host and validate the developed tools before deployment into a production environment.

We will present a generalized architecture that aims for a real-time, scalable, and extensible pipeline that can be deployed in diverse infrastructures to validate arbitrary attack detection tools. To motivate our approach, we will show an example deployment of our pipeline based on open-sourced tools. The example deployment uses as its data sources: (i) a customized honeypot environment at NCSA and (ii) a container-based testbed infrastructure for interactive attack replay. Each of these data sources is equipped with network and host-based monitoring tools such as Bro (a network-based intrusion detection system) and OSSEC (a host-based intrusion detection system) to allow for the runtime collection of data on system/user behavior. Finally, we will present an attack detection tool that we developed and that we look to validate through our pipeline. In conclusion, the talk will discuss the challenges of transitioning attack detection from theory to practice and how the proposed data analytics pipeline can help that transition.

This paper presents the architecture of an end-to-end secu- rity testbed and security analytics framework, which aims to: i) understand real-world exploitation of known security vulnerabilities and ii) preemptively detect multi-stage at- tacks, i.e., before the system misuse. With the increasing number of security vulnerabilities, it is necessary for secu- rity researchers and practitioners to understand: i) system and network behaviors under attacks and ii) potential ef- fects of attacks to the target infrastructure. To safely em- ulate and instrument exploits of known vulnerabilities, we use virtualization techniques to isolate attacks in contain- ers, e.g., Linux-based containers or Virtual Machines, and to deploy monitors, e.g., kernel probes or network packet captures, across a system and network stack. To infer the evolution of attack stages from monitoring data, we use a probabilistic graphical model, namely AttackTagger, that represents learned knowledge of simulated attacks in our se- curity testbed and real-world attacks. Experiments are be- ing run on a real-world deployment of the framework at the National Center for Supercomputing Applications (NCSA) at the University of Illinois at Urbana-Champaign.

Reliability and security tend to be treated separately because they appear orthogonal: reliability focuses on accidental failures, security on intentional attacks. Because of the apparent dissimilarity between the two, tools to detect and recover from different classes of failures and attacks are usually designed and implemented differently. So, integrating support for reliability and security in a single framework is a significant challenge.

Here, we discuss how to address this challenge in the context of cloud computing, for which reliability and security are growing concerns. Because cloud deployments usually consist of commodity hardware and software, efficient monitoring is key to achieving resiliency. Although reliability and security monitoring might use different types of analytics, the same sensing infrastructure can provide inputs to monitoring modules.

We split monitoring into two phases: logging and auditing. Logging captures data or events; it constitutes the framework’s core and is common to all monitors. Auditing analyzes data or events; it’s implemented and operated independently by each monitor. To support a range of auditing policies, logging must capture a complete view, including both actions and states of target systems. It must also provide useful, trustworthy information regarding the captured view.

We applied these principles when designing HyperTap, a hypervisor-level monitoring framework for virtual machines (VMs). Unlike most VM-monitoring techniques, HyperTap employs hardware architectural invariants (hardware invariants, for short) to establish the root of trust for logging. Hardware invariants are properties defined and enforced by a hardware platform (for example, the x86 instruction set architecture). Additionally, HyperTap supports continuous, event-driven VM monitoring, which enables both capturing the system state and responding rapidly to actions of interest.

Hypervisor activity is designed to be hidden from guest Virtual Machines (VM) as well as external observers. In this paper, we demonstrate that this does not always occur. We present a method by which an external observer can learn sensitive information about hypervisor internals, such as VM scheduling or hypervisor-level monitoring schemes, by observing a VM. We refer to this capability as Hypervisor Introspection (HI).

HI can be viewed as the inverse process of the well-known Virtual Machine Introspection (VMI) technique. VMI is a technique to extract VMs’ internal state from the hypervi- sor, facilitating the implementation of reliability and security monitors[1]. Conversely, HI is a technique that allows VMs to autonomously extract hypervisor information. This capability enables a wide range of attacks, for example, learning a hypervisor’s properties (version, configuration, etc.), defeating hypervisor-level monitoring systems, and compromising the confidentiality of co-resident VMs. This paper focuses on the discovery of a channel to implement HI, and then leveraging that channel for a novel attack against traditional VMI.

In order to perform HI, there must be a method of extracting information from the hypervisor. Since this information is intentionally hidden from a VM, we make use of a side channel. When the hypervisor checks a VM using VMI, VM execution (e.g. network communication between a VM and a remote system) must pause. Therefore, information regarding the hypervisor’s activity can be leaked through this suspension of execution. We call this side channel the VM suspend side channel, illustrated in Fig. 1. As a proof of concept, this paper presents how correlating the results of in-VM micro- benchmarking and out-of-VM reference monitoring can be used to determine when hypervisor-level monitoring tools are vulnerable to attacks.

This paper presents a solution that simultaneously addresses both reliability and security (RnS) in a monitoring framework. We identify the commonalities between reliability and security to guide the design of HyperTap, a hypervisor-level framework that efficiently supports both types of monitoring in virtualization environments. In HyperTap, the logging of system events and states is common across monitors and constitutes the core of the framework. The audit phase of each monitor is implemented and operated independently. In addition, HyperTap relies on hardware invariants to provide a strongly isolated root of trust. HyperTap uses active monitoring, which can be adapted to enforce a wide spectrum of RnS policies. We validate Hy- perTap by introducing three example monitors: Guest OS Hang Detection (GOSHD), Hidden RootKit Detection (HRKD), and Privilege Escalation Detection (PED). Our experiments with fault injection and real rootkits/exploits demonstrate that HyperTap provides robust monitoring with low performance overhead.

When SCADA systems are exposed to public networks, attackers can more easily penetrate the control systems that operate electrical power grids, water plants, and other critical infrastructures. To detect such attacks, SCADA systems require an intrusion detection technique that can understand the information carried by their usually proprietary network protocols.

To achieve that goal, we propose to attach to SCADA systems a specification-based intrusion detection framework based on Bro [7][8], a runtime network traffic analyzer. We have built a parser in Bro to support DNP3, a network protocol widely used in SCADA systems that operate electrical power grids. This built-in parser provides a clear view of all network events related to SCADA systems. Consequently, security policies to analyze SCADA-specific semantics related to the network events can be accurately defined. As a proof of concept, we specify a protocol validation policy to verify that the semantics of the data extracted from network packets conform to protocol definitions. We performed an experimental evaluation to study the processing capabilities of the proposed intrusion detection framework.

In the current generation of SCADA (Supervisory Control And Data Acquisition) systems used in power grids, a sophisticated attacker can exploit system vulnerabilities and use a legitimate maliciously crafted command to cause a wide range of system changes that traditional contingency analysis does not consider and remedial action schemes cannot handle. To detect such malicious commands, we propose a semantic analysis framework based on a distributed network of intrusion detection systems (IDSes). The framework combines system knowledge of both cyber and physical infrastructure in power grid to help IDS to estimate execution consequences of control commands, thus to reveal attacker’s malicious intentions. We evaluated the approach on the IEEE 30-bus system. Our experiments demonstrate that: (i) by opening 3 transmission lines, an attacker can avoid detection by the traditional contingency analysis and instantly put the tested 30-bus system into an insecure state and (ii) the semantic analysis provides reliable detection of malicious commands with a small amount of analysis time.