Search NASA⌕ Search

Engineering topics

Weber, Brian

Publications and source records attributed to Weber, Brian.

Towards a High Fidelity Training Environment for Autonomous Cyber Defense Agents

Cyber defenders are overwhelmed by the frequency and scale of attacks against their networks. This problem will only be exacerbated as attackers leverage AI to automate their workflows. Autonomous cyber defense capabilities could aid defenders by automating operations and adapting dynamically to novel threats. However, existing training environments fall short in areas such as generalization, explainability, scalability, and transferability, making it intractable to train agents that will be effective in real networks. In this paper we take an important step towards creating autonomous cyber defense agents — we present a high fidelity training environment called Cyberwheel that includes both simulation and emulation capabilities. Cyberwheel simplifies customization of the training network and easily allows redefining the agent’s reward function, observation space, and action space to support rapid experimentation of novel approaches to agent design. It also provides visibility into agent behaviors necessary for agent evaluation and sufficient documentation / examples to lower the barrier to entry. As an example use case of Cyberwheel, we present initial results training an autonomous agent to deploy cyber deception strategies in simulation.

Oesch, T↗

The Path to Autonomous Cyberdefense

Defenders are overwhelmed by attacks against their networks, which will only be exacerbated as attackers leverage artificial intelligence to automate workflows. Here, we propose a path to autonomous cyberagents able to augment defenders by automating critical steps in the cyberdefense lifecycle.

Oesch, Sean↗

AI ATAC 1: An Evaluation of Prominent Commercial Malware Detectors

This work presents an evaluation of six prominent commercial endpoint malware detectors, a network malware detector, and a file-conviction algorithm from a cyber technology vendor. The evaluation was administered as the first of the Artificial I ntelligence Applications t o Autonomous Cybersecurity (AI ATAC) prize challenges, funded by / completed in service of the US Navy. The experiment employed 100K files (50/50% benign/malicious) with a stratified distribution of file types, including ~1K zero-day program executables (increasing experiment size two orders of magnitude over previous work). We present an evaluation process of delivering a file to a fresh virtual machine donning the detection technology, waiting 90s to allow static detection, then executing the file and waiting another period for dynamic detection; this allows greater fidelity in the observational data than previous experiments, in particular, resource and time-to-detection statistics. To execute all 800K trials (100K files × 8 tools), a software framework is designed to choreograph the experiment into an automated, time-synced, and reproducible workflow with substantial parallelization. Software with base classes for this framework are provided. A cost-benefit model was configured to integrate the tools’ detection statistics into a comparable quantity by simulating costs of use. This provides a ranking methodology for cyber competitions and a lens for reasoning about the varied statistical results. The results provide insights on state of commercial malware detection.

Bridges, Robert↗

Virtual supervisory control and data acquisition (SCADA) automation controller

One or more virtual supervisory control and data acquisition (SCADA) controllers are provided. The virtual SCADA controller(s) may be deployed upon a detection of a failure of a physical SCADA controller. An on-site module may detect failure including a security breach which causes the physical SCADA controller to be disconnected from a network and the virtual SCADA controller(s) to take over control. The virtual SCADA controller(s) may be on-site or off-site.

Smith, Jared M.↗

Testing SOAR tools in use

Investigations within Security Operation Centers (SOCs) are tedious as they rely on manual efforts to query diverse data sources, overlay related logs, correlate the data into information, and then document results in a ticketing system. Security Orchestration, Automation, and Response (SOAR) tools are a relatively new technology that promise, with appropriate configuration, to collect, filter, and display needed diverse information; automate many of the common tasks that unnecessarily require SOC analysts’ time; facilitate SOC collaboration; and, in doing so, improve both efficiency and consistency of SOCs. There has been no prior research to test SOAR tools in practice; hence, understanding and evaluation of their effect is nascent and needed. Here, in this paper, we design and administer the first hands-on user study of SOAR tools, involving 24 participants and six commercial SOAR tools. Our contributions include the experimental design, itemizing six characteristics of SOAR tools, and a methodology for testing them. We describe configuration of a cyber range test environment, including network, user, and threat emulation; a full SOC tool suite; and creation of artifacts allowing multiple representative investigation scenarios to permit testing. We present the first research results on SOAR tools. Concisely, our findings are that: per-SOC SOAR configuration is extremely important; SOAR tools increase efficiency and reduce context switching, although with potentially decreased ticketing accuracy/completeness; user preference is slightly negatively correlated with their performance with the tool; internet dependence varies widely among SOAR tools; and balance of automation with assisting decision making is preferred by senior participants. We deliver a public user- and tool-anonymized and -obfuscated version of the data.

97 MATHEMATICS AND COMPUTING↗

Beyond the Hype: An Evaluation of Commercially Available Machine-Learning-Based Malware Detectors

There is a lack of scientific testing of commercially available malware detectors, especially those that boast accurate classification of never-before-seen (i.e., zero-day) files using machine learning (ML). Consequently, efficacy of malware detectors is opaque, inhibiting end users from making informed decisions and researchers from targeting gaps in current detectors. In this paper, we present a scientific evaluation of four prominent commercial malware detection tools to assist an organization with two primary questions: To what extent do ML-based tools accurately classify previously and never-before-seen files? Is purchasing a network-level malware detector worth the cost? To investigate, we tested each tool against 3,536 total files (2,554 or 72% malicious, 982 or 28% benign) of a variety of file types, including hundreds of malicious zero-days, polyglots, and APT-style files, delivered on multiple protocols. We present statistical results on detection time and accuracy, consider complementary analysis (using multiple tools together), and provide two novel applications of the recent cost-benefit evaluation procedure of Iannacone & Bridges. Although the ML-based tools are more effective at detecting zero-day files and executables, the signature-based tool might still be an overall better option. Both network-based tools provide substantial (simulated) savings when paired with either host tool, yet both show poor detection rates on protocols other than HTTP or SMTP. Our results show that all four tools have near-perfect precision but alarmingly low recall, especially on file types other than executables and office files—37% of malware, including all polyglot files, were undetected. Priorities for researchers and takeaways for end users are given. Code for future use of the cost model is provided.

97 MATHEMATICS AND COMPUTING↗

Toward the Detection of Polyglot Files

Standardized file types play a key role in the development and use of computer software. However, it is possible to confound standardized file type processing by creating a file that is valid in multiple file types. The resulting polyglot (many languages) file can confuse file type identification, allowing elements of the file to evade analysis. This is especially problematic for malware detection systems that rely on file type identification for feature extraction. Although work has been done to identify file types using more comprehensive methods than file signatures, accurate identification of polyglot files remains an open problem. Since malware detection systems routinely perform file type-specific feature extraction, polyglot files need to be filtered out prior to ingestion by these systems. Otherwise, malicious content could pass through undetected. To address the problem of polyglot detection we assembled a data set using the mitra tool. We then evaluated the performance of the most commonly used file identification tools, including file, polydet, binwalk, and TrID. Our analysis demonstrates that existing file type detection tools fail to provide reliable polyglot detection. We then evaluated the ability of a range of machine and deep learning models to detect polyglot files. The most performant models were MalConv2 and Catboost, which demonstrated the highest recall on our data set with 95.16% and 95.45%, respectively. These models outperformed existing methods and could be incorporated into a malware detector’s file processing pipeline to filter out potentially malicious polyglots before file type-dependent feature extraction takes place.

Koch, Luke↗

VAC: A Software Approach to Resilient SCADA Automation

To better secure critical infrastructure, especially power systems, this paper introduces a virtual SCADA automation controller. The automation controller is a gateway into a power subsystem, making it a valuable target for cyber-attacks that could cut it off from the control center and cause a loss of view and control. To prevent this, the Virtual Automation Controller (VAC) is a backup device that mirrors the capabilities of the physical controller. It can communicate via Modbus and DNP3 and is containerized so it can be deployed on a variety of platforms. Furthermore, it utilizes software-defined networking to quickly disconnect a failed automation controller and preserve its state for forensics. The VAC gives system operators time to replace the failed controller and prevents dangerous and costly damage to power systems. The VAC is compared against the SEL 3505-3 RTAC and shown to have the necessary features to act as a failover controller.

Johnson, Jordan↗

Soar Survey Scoring

ORNL conducted an experiment to test COTS (commercial off the shelf) SOAR tools. Vendors interested in entering their tool into the experiment were asked to submit information videos about their tool for phase 1 to down-select which tools should make it through to the testing phase. SOC operators were assigned to watch a subset of the videos such that each video was assigned to the same number of operators. After watching all the videos, operators completed surveys which asked them to grade basic aspects about the tool as best as they could based on the provided video, and asked them to rank the order of their interest in the vendors they were assigned to watch videos on. The problem solved was how do we aggregate the results of the surveys and each of the operator's ranked lists into an overall ranking of all the tools. In addition, it was desirable to be able to work on this code before final results were collected from operators, so code to generate fake data was developed to output in the same output as the survey results.

Bridges, Robert↗

pyvnc_client

A python VNC client written from scratch using python's socket library

Weber, Brian↗

D2U: Data Driven User Emulation for the Enhancement of Cyber Testing, Training, and Data Set Generation

Whether testing intrusion detection systems, conducting training exercises, or creating data sets to be used by the broader cybersecurity community, realistic user behavior is a critical component of a cyber range. Existing methods either rely on network level data or replay recorded user actions to approximate real users in a network. Our work is the first to produce generative models trained on actual user data (sequences of application usage) collected on endpoints. Once trained to the user's behavioral data, these models can generate novel sequences of actions %that appear to come from the same distribution as the training data. These sequences of actions are then fed to our custom software via configuration files, which replicate those behaviors on end devices. Notably, our models are platform agnostic and could generate behavior data for any emulation software package. In this paper we present our model generation process, software architecture, and an initial evaluation of the fidelity of our models. Our software is currently deployed in a cyber range to help evaluate the efficacy of defensive cyber technologies. We suggest additional ways that the cyber community as a whole can benefit from more realistic user behavior emulation. The data used to train our model, as well as sample configuration files produced by the model, are available at [redacted].

Oesch, T↗