DOCSLIB.ORG

Tracking Rootkit Footprints with a Practical Memory Analysis System

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Tracking Rootkit Footprints with a Practical Memory Analysis System Tracking Rootkit Footprints with a Practical Memory Analysis System Weidong Cui Marcus Peinado Microsoft Research Microsoft Research [email protected] [email protected] Zhilei Xu Ellick Chan Massachusetts Institute of Technology University of Illinois at Urbana-Champaign [email protected] [email protected] Abstract An important task in detecting and analyzing kernel rootkits is to identify all the changes a rootkit makes to In this paper, we present MAS, a practical memory anal- an infected OS kernel for hijacking code execution or ysis system for identifying a kernel rootkit’s memory hiding its activities. We call these changes a rootkit’s footprint in an infected system. We also present two memory footprint. We perform this task in two common large-scale studies of applying MAS to 848 real-world scenarios: We detect if real-world computer systems are Windows kernel crash dumps and 154,768 potential mal- infected by kernel rootkits. We also analyze suspicious ware samples. software in a controlled environment. One can use either Error propagation and invalid pointers are two key execution tracing or memory analysis in a controlled en- challenges that stop previous pointer-based memory vironment, but is usually limited to memory analysis for traversal solutions from effectively and efficiently ana- real-world systems. In this paper we focus on the mem- lyzing real-world systems. MAS uses a new memory ory analysis approach since it can be applied in both sce- traversal algorithm to support error correction and stop narios. error propagation. Our enhanced static analysis allows After many years of research on kernel rootkits, we the MAS memory traversal to avoid error-prone opera- still lack a practical memory analysis system that is ac- tions and provides it with a reliable partial type assign- curate, robust, and performant. In other words, we ex- ment. pect such a practical system to correctly and quickly Our experiments show that MAS was able to analyze identify all memory changes made by a rootkit to arbi- all memory snapshots quickly with typical running times trary systems that may have a variety of kernel modules between 30 and 160 seconds per snapshot and with near loaded. Furthermore, we lack a large-scale study of ker- perfect accuracy. Our kernel malware study observes nel rootkit behaviors, partly because there is no practi- that the malware samples we tested hooked 191 differ- cal system that can analyze memory infected by kernel ent function pointers in 31 different data structures. With rootkits in an accurate, robust and performant manner. MAS, we were able to determine quickly that 95 out of In this paper, we present MAS, a practical memory the 848 crash dumps contained kernel rootkits. analysis system for identifying a rootkit’s memory foot- print. We also present the results of two large-scale ex- 1 Introduction periments in which we use MAS to analyze 837 kernel crash dumps of real-world systems running Windows 7, Kernel rootkits represent a significant threat to computer and 154,768 potential malware samples from the reposi- security because, once a rootkit compromises the OS ker- tory of a major commercial anti-malware vendor. These nel, it owns the entire software stack which allows it to are the two major contributions of this paper. evade detections and launch many kinds of attacks. For Previous work [2, 3, 19] has established that, to iden- instance, the Alureon rootkit [1] was infamous for steal- tify a rootkit’s memory footprint, we need to check not ing passwords and credit card data, running botnets, and only the integrity of kernel code and static data but also causing a large number of Windows systems to crash. the integrity of dynamic data, and the real challenge lies Kernel rootkits also present a serious challenge for mal- in the latter task. ware analysis because, to hide its existence, a rootkit at- In order to locate dynamic data, these systems first lo- tempts to manipulate the kernel code and data of an in- cate static data objects in each loaded module, then re- fected system. cursively follow the pointers in these objects and in all newly identified data objects, until no new data object crash dumps taken from 837 real-world systems run- can be added. Unlike the earlier systems, KOP [3] in- ning Windows 7 and memory snapshots taken from Win- cludes generic pointers (e.g., void∗) in its memory traver- dows XP SP3 VMs subjected to one of 154,768 potential sal, and shows that failing to do so will prevent the mem- real-world malware samples. For the Windows 7 crash ory traversal from reaching about two thirds of the dy- dumps, MAS took 105 seconds to analyze a single dump namic objects. on average. It identified a total of about 400,000 suspi- Previous solutions do not sufficiently address an im- cious function pointers. We were able to verify the cor- portant practical problem of this memory traversal pro- rectness of all but 24 of them. Moreover, with the results cedure: its tendency to accumulate and propagate errors. of MAS, we were able to quickly identify 90 Windows 7 A typical large real-world kernel memory image is bound crash dumps (and five Windows Vista SP1 crash dumps) to contain invalid pointers. That is, there are likely to be that were infected by kernel rootkits. In our study of dynamic objects with pointer fields not pointing to valid malware samples, MAS required about 30 seconds to objects. Following such pointers results in objects being analyze each VM memory snapshot. Our study shows incorrectly included in the object mapping. Worse, such that the kernel rootkits we tested hooked 191 function identification errors can be propagated due to the nature pointer fields in 31 data structures. It also shows that of the recursive, greedy memory traversal. A single in- many malware samples had identical footprints, which correctly identified data object may cause many more suggests that we can use MAS to detect new malware mistakes in the subsequent traversal. samples/families that have different memory footprints. Invalid pointers may exist for a variety of reasons. For The rest of this paper is organized as follows. Sec- example, an object may have been allocated, but not yet tion 2 provides an overview of the paper. Sections 3 and initialized. KOP is exposed to a second source of poten- 4 describe the design of MAS and explain the algorithms tial errors. KOP tries to follow all generic pointers. If the used for static analysis and memory traversal. Section 5 pointer type cannot be uniquely determined, KOP tries explains how we evaluate the set of objects found by to decide the correct type using a heuristic. A fraction of MAS for suspicious activity. Section 6 describes our these guesses are bound to be incorrect. implementation of MAS. Section 7 describes our evalua- In light of these problems, we design MAS to con- tion of MAS. Section 8 and Section 9 describe two large- trol the number of errors that arise from following invalid scale experiments in which we analyze malware samples pointers and to contain their effects. Instead of perform- and identify rootkits from crash dumps. Sections 10 and ing a greedy memory traversal that is vulnerable to error Section 11 discuss related work and limitations. Finally, propagation, MAS uses a new traversal scheme to sup- Section 12 concludes the paper. port error correction. MAS also uses static analysis to derive information that can be used to uniquely identify 2 Overview many objects and their types without having to rely on the recursive traversal procedure. Furthermore, MAS is The goal of MAS is to identify all memory changes a not subject to errors caused by ambiguous pointers, i.e., rootkit makes for hijacking execution and hiding its ac- pointers whose type cannot be uniquely determined. It tivities. MAS does so in three steps: static analysis, uses an enhanced static analysis to identify unique types memory traversal, and integrity checking. for a large fraction of generic pointers and ignores all remaining ambiguous pointers. While this may reduce Static Analysis: MAS takes the source code of the OS coverage, it will never cause an object to be recognized kernel and drivers as the input and uses a pointer incorrectly. Our evaluation will show that the impact on analysis algorithm to identify candidate types for coverage is minor. Finally, before accepting an object, generic pointers such as void∗ and linked list con- MAS checks a number of constraints, including new con- structs. Furthermore, it also computes the associa- straints we derive from our static analysis. tions between data types and pool tags [18]. We implemented a prototype of MAS and compared Memory Traversal: MAS tries to identify dynamic it with KOP on eleven crash dumps of real-world sys- data objects in a given memory snapshot. Besides tems running Windows Vista SP1. MAS’s performance the snapshot, the input includes the type related in- is one order of magnitude better than KOP regarding both formation derived from static analysis and the sym- static analysis and memory traversal. MAS did not miss bol information [15] for each loaded module (if it is or misidentify any function pointers found by KOP, but available). KOP missed or misidentified up to 40% of suspicious function pointers (i.e., function pointers that point to un- Integrity Checking: MAS identifies the memory trusted code). changes a rootkit makes by inspecting the integrity In our large-scale experiments, we ran MAS over of code, static data and dynamic data (recognized 2 from memory traversal). In addition to checking #"$"%&'" !" !" +" if some code section is modified, MAS detects two %&" !" &" !" (&" +" *" (#"$"%)'" kinds of violations: (1) a function pointer points to +" !" !" +" (&"$"*'" +" +" a memory region outside of a list of known good !" +" !" modules; (2) a data object is hidden from a system !")"$"*" #" !" (#" +" %)" !" )" program.
Load more

Recommended publications
  • A the Hacker
    A The Hacker Madame Curie once said “En science, nous devons nous int´eresser aux choses, non aux personnes [In science, we should be interested in things, not in people].” Things, however, have since changed, and today we have to be interested not just in the facts of computer security and crime, but in the people who perpetrate these acts. Hence this discussion of hackers. Over the centuries, the term “hacker” has referred to various activities. We are familiar with usages such as “a carpenter hacking wood with an ax” and “a butcher hacking meat with a cleaver,” but it seems that the modern, computer-related form of this term originated in the many pranks and practi- cal jokes perpetrated by students at MIT in the 1960s. As an example of the many meanings assigned to this term, see [Schneier 04] which, among much other information, explains why Galileo was a hacker but Aristotle wasn’t. A hack is a person lacking talent or ability, as in a “hack writer.” Hack as a verb is used in contexts such as “hack the media,” “hack your brain,” and “hack your reputation.” Recently, it has also come to mean either a kludge, or the opposite of a kludge, as in a clever or elegant solution to a difficult problem. A hack also means a simple but often inelegant solution or technique. The following tentative definitions are quoted from the jargon file ([jargon 04], edited by Eric S. Raymond): 1. A person who enjoys exploring the details of programmable systems and how to stretch their capabilities, as opposed to most users, who prefer to learn only the minimum necessary.
    [Show full text]
  • Diapositiva 1
    Feliciano Intini Responsabile dei programmi di Sicurezza e Privacy di Microsoft Italia • NonSoloSecurity Blog: http://blogs.technet.com/feliciano_intini • Twitter: http://twitter.com/felicianointini 1. Introduction - Microsoft Security Intelligence Report (SIR) 2. Today‘s Threats - SIR v.8 New Findings – Italy view 3. Advancements in Software Protection and Development 4. What the Users and Industry Can Do The 8th volume of the Security Intelligence Report contains data and intelligence from the past several years, but focuses on the second half of 2009 (2H09) Full document covers Malicious Software & Potentially Unwanted Software Email, Spam & Phishing Threats Focus sections on: Malware and signed code Threat combinations Malicious Web sites Software Vulnerability Exploits Browser-based exploits Office document exploits Drive-by download attacks Security and privacy breaches Software Vulnerability Disclosures Microsoft Security Bulletins Exploitability Index Usage trends for Windows Update and Microsoft Update Microsoft Malware Protection Center (MMPC) Microsoft Security Response Center (MSRC) Microsoft Security Engineering Center (MSEC) Guidance, advice and strategies Detailed strategies, mitigations and countermeasures Fully revised and updated Guidance on protecting networks, systems and people Microsoft IT ‗real world‘ experience How Microsoft IT secures Microsoft Malware patterns around the world with deep-dive content on 26 countries and regions Data sources Malicious Software and Potentially Unwanted Software MSRT has a user base
    [Show full text]
  • Proposal of N-Gram Based Algorithm for Malware Classification
    SECURWARE 2011 : The Fifth International Conference on Emerging Security Information, Systems and Technologies Proposal of n-gram Based Algorithm for Malware Classification Abdurrahman Pektaş Mehmet Eriş Tankut Acarman The Scientific and Technological The Scientific and Technological Computer Engineering Dept. Research Council of Turkey Research Council of Turkey Galatasaray University National Research Institute of National Research Institute of Istanbul, Turkey Electronics and Cryptology Electronics and Cryptology e-mail:[email protected] Gebze, Turkey Gebze, Turkey e-mail:[email protected] e-mail:[email protected] Abstract— Obfuscation techniques degrade the n-gram malware used by the previous work. Similar methodologies features of binary form of the malware. In this study, have been used in source authorship, information retrieval methodology to classify malware instances by using n-gram and natural language processing [5], [6]. features of its disassembled code is presented. The presented The first known use of machine learning in malware statistical method uses the n-gram features of the malware to detection is presented by the work of Tesauro et al. in [7]. classify its instance with respect to their families. n-gram is a fixed size sliding window of byte array, where n is the size of This detection algorithm was successfully implemented in the window. The contribution of the presented method is IBM’s antivirus scanner. They used 3-grams as a feature set capability of using only one vector to represent malware and neural networks as a classification model. When the 3- subfamily which is called subfamily centroid. Using only one grams parameter is selected, the number of all n-gram vector for classification simply reduces the dimension of the n- features becomes 2563, which leads to some spacing gram space.
    [Show full text]
  • Download Slides
    Scott Wu Point in time cleaning vs. RTP MSRT vs. Microsoft Security Essentials Threat events & impacts More on MSRT / Security Essentials MSRT Microsoft Windows Malicious Software Removal Tool Deployed to Windows Update, etc. monthly since 2005 On-demand scan on prevalent malware Microsoft Security Essentials Full AV RTP Inception in Oct 2009 RTP is the solution One-off cleaner has its role Quiikck response Workaround Baseline ecosystem cleaning Industrypy response & collaboration Threat Events Worms (some are bots) have longer lifespans Rogues move on quicker MarMar 2010 2010 Apr Apr 2010 2010 May May 2010 2010 Jun Jun 2010 2010 Jul Jul 2010 2010 Aug Aug 2010 2010 1,237,15 FrethogFrethog 979,427 979,427 Frethog Frethog 880,246880,246 Frethog Frethog465,351 TaterfTaterf 5 1,237,155Taterf Taterf 797,935797,935 TaterfTaterf 451,561451,561 TaterfTaterf 497,582 497,582 Taterf Taterf 393,729393,729 Taterf Taterf447,849 FrethogFrethog 535,627535,627 AlureonAlureon 493,150 493,150 AlureonAlureon 436,566 436,566 RimecudRimecud 371,646 371,646 Alureon Alureon 308,673308,673 Alureon Alureon 441,722 RimecudRimecud 341,778341,778 FrethogFrethog 473,996473,996 BubnixBubnix 348,120 348,120 HamweqHamweq 289,603 289,603 Rimecud Rimecud289,629 289,629 Rimecud Rimecud318,041 AlureonAlureon 292,810 292,810 BubnixBubnix 471,243 471,243 RimecudRimecud 287,942287,942 ConfickerConficker 286,091286, 091 Hamwe Hamweqq 250,286250, 286 Conficker Conficker220,475220, 475 ConfickerConficker 237237,348, 348 RimecudRimecud 280280,440, 440 VobfusVobfus 251251,335, 335
    [Show full text]
  • Classification of Malware Models Akriti Sethi San Jose State University
    San Jose State University SJSU ScholarWorks Master's Projects Master's Theses and Graduate Research Spring 5-20-2019 Classification of Malware Models Akriti Sethi San Jose State University Follow this and additional works at: https://scholarworks.sjsu.edu/etd_projects Part of the Artificial Intelligence and Robotics Commons, and the Information Security Commons Recommended Citation Sethi, Akriti, "Classification of Malware Models" (2019). Master's Projects. 703. DOI: https://doi.org/10.31979/etd.mrqp-sur9 https://scholarworks.sjsu.edu/etd_projects/703 This Master's Project is brought to you for free and open access by the Master's Theses and Graduate Research at SJSU ScholarWorks. It has been accepted for inclusion in Master's Projects by an authorized administrator of SJSU ScholarWorks. For more information, please contact [email protected]. Classification of Malware Models A Project Presented to The Faculty of the Department of Computer Science San José State University In Partial Fulfillment of the Requirements for the Degree Master of Science by Akriti Sethi May 2019 © 2019 Akriti Sethi ALL RIGHTS RESERVED The Designated Project Committee Approves the Project Titled Classification of Malware Models by Akriti Sethi APPROVED FOR THE DEPARTMENT OF COMPUTER SCIENCE SAN JOSÉ STATE UNIVERSITY May 2019 Dr. Mark Stamp Department of Computer Science Dr. Thomas Austin Department of Computer Science Dr. Philip Heller Department of Computer Science ABSTRACT Classification of Malware Models by Akriti Sethi Automatically classifying similar malware families is a challenging problem. In this research, we attempt to classify malware families by applying machine learning to machine learning models. Specifically, we train hidden Markov models (HMM) for each malware family in our dataset.
    [Show full text]
  • |||FREE||| Meltdown
    MELTDOWN FREE DOWNLOAD Robert Rigby,Andy McNab | 304 pages | 01 May 2008 | Random House Children's Publishers UK | 9780552552240 | English | London, United Kingdom Meltdown and Spectre Most certainly, yes. Eee-o eleven Meltdown and Spectre Vulnerabilities in Meltdown computers leak passwords and sensitive data. Meltdown could Meltdown impact a wider Meltdown of computers than presently identified, as there is little to no variation in the microprocessor families used by Meltdown computers. Help Learn to edit Community portal Recent changes Upload file. In fact, the safety checks of said best practices actually increase the attack surface and may make applications more susceptible to Spectre. What can be leaked? Vox Media, Inc. We would Meltdown to thank Intel for awarding us with a bug bounty for the responsible disclosure process, and their professional handling of this issue through communicating a clear timeline and connecting all involved researchers. Mitigation of the vulnerability requires changes to operating Meltdown kernel code, including increased isolation of kernel memory from user-mode Meltdown. DarkSeoul CryptoLocker. Send us feedback. Life - reality at large- becomes overwhelming. Rush B Cyka Meltdown Play the game. Meltdown Note. Logos are designed by Natascha Eibl. Who reported Spectre? English Language Learners Definition of meltdown. Cloud providers Meltdown users to execute programs on the same physical servers where sensitive data might be stored, and rely Meltdown safeguards provided by the Meltdown to prevent unauthorized access to the privileged memory locations where that data is stored, a feature that the Meltdown exploit circumvents. Categories Meltdown Computer security exploits Hardware bugs Intel x86 microprocessors Side-channel attacks Speculative execution security vulnerabilities in computing X86 architecture X86 memory management.
    [Show full text]
  • The Evolution of TDL: Conquering X64
    The Evolution of TDL: Conquering x64 Revision 1.1 Eugene Rodionov, Malware Researcher Aleksandr Matrosov, Senior Malware Researcher 2 2 CONTENTS 3 INTRODUCTION ..................................................................................................................................................... 4 1 INVESTIGATION ............................................................................................................................................. 5 1.1 GANGSTABUCKS ............................................................................................................................................... 6 2 INSTALLATION ............................................................................................................................................. 11 2.1 INFECTING X86 SYSTEMS .................................................................................................................................. 11 2.2 INFECTING X64 SYSTEMS .................................................................................................................................. 13 2.3 THE DROPPER’S PAYLOAD ................................................................................................................................ 14 2.4 COMPARISON WITH TDL3/TDL3+..................................................................................................................... 15 3 THE BOT .....................................................................................................................................................
    [Show full text]
  • A Threat Modeling Framework for Evaluating Computing Platforms Against Architectural Attacks
    A Threat Modeling Framework for Evaluating Computing Platforms Against Architectural Attacks Seyyedeh Atefeh Musavi1 and Mahmoud Reza Hashemi1 School of Electrical and Computer Engineering, College of Engineering, University of Tehran, Tehran, Iran. amusavi,[email protected] Abstract. Architectural attacks are kinds of hardware-involved software attacks in which a software component misuse a privileged relationship with the hardware to by pass system protections, monitors, or forensic tools. These relationships are often not illegal and exist between system components by design. Hence, even a system with secure hardware and software components, can be architecturally vulnerable. Unfortunately, the existing threat modeling schemes are not applicable for modeling architectural attacks against computing platforms. This is mostly because the existing techniques rely on an abstract representation of a software (.e.g., Data Flow Diagram) as a primary requirement which is not available for a platform as a whole (considering both hardware and software elements). In this paper, we have discussed the necessity of a hardware-software architectural view to system threat modeling. Then, we have proposed Lamellae, a framework adapts threat modeling method to be applicable for untrusted platforms by a holistic approach. Lamellae involves system security architecture for abstract modeling of the platforms. Using the Design structure matrix analysis, Lamellae helps an end-user to identify possible attack vectors against a platform. The framework is a connection point of concepts from system engineering and software security domains. We have applied the framework on a multi-purpose computer with x86-64 architecture as a case-study to show the effectiveness of our framework.
    [Show full text]
  • Microsoft | Security Intelligence Report
    Battling Botnets for Control of Computers Microsoft | Security Intelligence Report Volume 9 January through June 2010 Microsoft | Security Intelligence Report Microsoft Security Intelligence Report This document is for informational purposes only. MICROSOFT MAKES NO WARRANTIES, EXPRESS, IMPLIED, OR STATUTORY, AS TO THE INFORMA- TION IN THIS DOCUMENT. This document is provided “as-is.” Information and views expressed in this document, including URL and other Internet Web site references, may change without notice. You bear the risk of using it. Copyright © 2010 Microsoft Corporation. All rights reserved. The names of actual companies and products mentioned herein may be the trademarks of their respective owners. 2 January through June 2010 Authors David Anselmi Jimmy Kuo Navaneethan Santhanam Digital Crimes Unit Microsoft Malware Protection Center Bing Richard Boscovich Scott Molenkamp Christian Seifert Digital Crimes Unit Microsoft Malware Protection Center Bing T.J. Campana Michelle Meyer Frank Simorjay Digital Crimes Unit Microsoft Trustworthy Computing Microsoft Trustworthy Computing Neil Carpenter Bala Neerumalla Holly Stewart CSS Security Microsoft Secure SQL Initiative Team Microsoft Malware Protection Center Greg Cottingham Daryl Pecelj Adrian Stone CSS Security Microsoft IT Information Security and Risk Management Microsoft Security Response Center Joe Faulhaber Anthony Penta Matt Thomlinson Microsoft Malware Protection Center Microsoft Windows Safety Platform Microsoft Security Response Center Vinny Gullotto Paul Pottorff Jossie
    [Show full text]
  • Cyber Crime a Technical Desk Reference for Additional Copies Or to Download This Document, Please Visit
    Cyber Crime A Technical Desk Reference For additional copies or to download this document, please visit: http://msisac.cisecurity.org/resources/guides © 2013 Center for Internet Security. All rights reserved. The information in this document is provided by the Multi-State Information Sharing and Analysis Center (MS-ISAC), a division of the Center for Internet Security (CIS), for non-commerical informational and educational purposes only. CIS does not warrant the accuracy or completeness of the information or commit to issue updates or corrections to the information. CIS is not responsible for any damages resulting from use of or reliance on the information contained herein. The mission of the Center for Internet Security is to enhance the security readiness and response of public and private sector entities, with a commitment to excellence through collaboration. CIS comprises four program divisions: Integrated Intelligence Center The mission of the Integrated Intelligence Center is to facilitate trusted relationships with government and private sector entities to develop and disseminate comprehensive, coordinated intelligence products that help improve the security posture of all partners. Multi-State Information Sharing and Analysis Center The mission of the MS-ISAC is to improve the overall cyber security posture of state, local, territorial and tribal governments. Collaboration and information sharing among members, private sector partners and the U.S. Department of Homeland Security are the keys to success. Security Benchmarks The mission of the division is to establish and promote the use of consensus- based best practice standards to raise the level of security and privacy in Internet-connected systems, and to ensure the integrity of the public and private Internet-based functions and transactions on which society increasingly depends.
    [Show full text]
  • Cyber Crime a Technical Desk Reference for Additional Copies Or to Download This Document, Please Visit
    Cyber Crime A Technical Desk Reference For additional copies or to download this document, please visit: http://msisac.cisecurity.org/resources/guides © 2013 Center for Internet Security. All rights reserved. The information in this document is provided by the Multi-State Information Sharing and Analysis Center (MS-ISAC), a division of the Center for Internet Security (CIS), for non-commerical informational and educational purposes only. CIS does not warrant the accuracy or completeness of the information or commit to issue updates or corrections to the information. CIS is not responsible for any damages resulting from use of or reliance on the information contained herein. The mission of the Center for Internet Security is to enhance the security readiness and response of public and private sector entities, with a commitment to excellence through collaboration. CIS comprises four program divisions: Integrated Intelligence Center The mission of the Integrated Intelligence Center is to facilitate trusted relationships with government and private sector entities to develop and disseminate comprehensive, coordinated intelligence products that help improve the security posture of all partners. Multi-State Information Sharing and Analysis Center The mission of the MS-ISAC is to improve the overall cyber security posture of state, local, territorial and tribal governments. Collaboration and information sharing among members, private sector partners and the U.S. Department of Homeland Security are the keys to success. Security Benchmarks The mission of the division is to establish and promote the use of consensus- based best practice standards to raise the level of security and privacy in Internet-connected systems, and to ensure the integrity of the public and private Internet-based functions and transactions on which society increasingly depends.
    [Show full text]
  • Survey and Taxonomy of Botnet Research Through Life-Cycle
    Survey and Taxonomy of Botnet Research through Life-Cycle RAFAEL A. RODR´IGUEZ-GOMEZ,´ GABRIEL MACIA-FERN´ ANDEZ´ and PEDRO GARC´IA-TEODORO, University of Granada Of all current threats to cybersecurity, botnets are at the top of the list. In consequence, interest in this problem is increasing rapidly among the research community and the number of publications on the question has grown exponentially in recent years. This article proposes a taxonomy of botnet research and presents a survey of the field to provide a comprehensive overview of all these contributions. Furthermore, we hope to provide researchers with a clear perspective of the gaps that remain to be filled in our defenses against botnets. The taxonomy is based upon the botnet’s life-cycle, defined as the sequence of stages a botnet needs to pass through in order to reach its goal. This approach allows us to consider the problem of botnets from a global perspective, which constitutes a key difference from other taxonomies that have been proposed. Under this novel taxonomy, we conclude that all attempts to defeat botnets should be focused on one or more stages of this life-cycle. In fact, the sustained hindering of any of the stages makes it possible to thwart a botnet’s progress and thus render it useless. We test the potential capabilities of our taxonomy by means of a survey of current botnet research, and find it genuinely useful in understanding the focus of the different contributions in this field. Categories and Subject Descriptors: K.6.5 [Security and Protection]: Invasive Software; K.4.2 [Social Issues]: Abuse and Crime Involving Computers General Terms: Security Additional Key Words and Phrases: Attack, botnet, defense, detection, survey, taxonomy 45 ACM Reference Format: Rodr´ıguez-Gomez,´ R.
    [Show full text]