WOPR: A Dynamic Cybersecurity Detection and Response Framework

Loading...
Thumbnail Image

Authors

Walker, Aaron

Issue Date

2021

Type

Dissertation

Language

Keywords

Extreme Learning Machine , Incident Response , Machine Learning , Malware , Threat

Research Projects

Organizational Units

Journal Issue

Alternative Title

Abstract

Malware authors develop software to exploit the flaws in any platform and application which suffers a vulnerability in its defenses, be it through unpatched known attack vectors or zero-day attacks for which there is no current solution. It is the responsibility of cybersecurity personnel to monitor, detect, respond to and protect against such incidents that could affect their organization. Unfortunately, the low number of skilled, available cybersecurity professionals in the job market means that many positions go unfilled and cybersecurity threats are unknowingly allowed to negatively affect many enterprises.The demand for a greater cybersecurity posture has led several organizations to de- velop automated threat analysis tools which can be operated by less-skilled infor- mation security analysts and response teams. However, the diverse needs and organizational factors of most businesses presents a challenge for a “one size fits all” cybersecurity solution. Organizations in different industries may not have the same regulatory and standards compliance concerns due to processing different forms and classifications of data. As a result, many common security solutions are ill equipped to accurately model cybersecurity threats as they relate to each unique organization.We propose WOPR, a framework for automated static and dynamic analysis of software to identify malware threats, classify the nature of those threats, and deliver an appropriate automated incident response. Additionally, WOPR provides the end user the ability to adjust threat models to fit the risks relevant to an organization, allowing for bespoke automated cybersecurity threat management. Finally, WOPR presents a departure from traditional signature-based detection found in anti-virus and intrusion detection systems through learning system-level behavior and matching system calls with malicious behavior.

Description

Citation

Publisher

Journal

Volume

Issue

PubMed ID

DOI

ISSN

EISSN