Skip to content

Past selected articles

Archived past selected research articles in reverse chronological order.

May 2014

Forensic Zero-Knowledge Event Reconstruction on Filesystem Metadata, by Sven Kälber, Andreas Dewald, Steffen Idler, 2014

Abstract: Criminal investigations today can hardly be imagined without the forensic analysis of digital devices, regardless of whether it is a desktop computer, a mobile phone, or a navigation system. This not only holds true for cases of cybercrime, but also for traditional delicts such as murder or blackmail, and also private corporate investigations rely on digital forensics. This leads to an increasing number of cases with an ever-growing amount of data, that exceeds the capacity of the forensic experts. To support investigators to work more efficiently, we introduce a novel approach to automatically reconstruct events that previously occurred on the examined system and to provide a quick overview to the investigator as a starting point for further investigation. In contrast to the few existing approaches, our solution does not rely on any previously profiled system behavior or knowledge about specific applications, log files, or file formats. We further present a prototype implementation of our so-called zero knowledge event reconstruction approach, that solely tries to make sense of characteristic structures in file system metadata such as file- and folder-names and timestamps.

Measurement and Analysis of Child Pornography Trafficking on P2P Networks, by Hurley, Ryan, Prusty, Swagatika, Soroush, Hamed, Walls, Robert J., Albrecht, Jeannie, Cecchet, Emmanuel, Levine, Brian Neil, Liberatore, Marc, Lynn, Brian, Wolak, Janis, 2013.

Abstract: Peer-to-peer networks are the most popular mechanism for the criminal acquisition and distribution of child pornography (CP). In this paper, we examine observations of peers sharing known CP on the eMule and Gnutella networks, which were collected by law enforcement using forensic tools that we developed. We characterize a year's worth of network activity and evaluate different strategies for prioritizing investigators' limited resources. The highest impact research in criminal forensics works within, and is evaluated under, the constraints and goals of investigations. We follow that principle, rather than presenting a set of isolated, exploratory characterizations of users.

First, we focus on strategies for reducing the number of CP files available on the network by removing a minimal number of peers. We present a metric for peer removal that is more effective than simply selecting peers with the largest libraries or the most days online. Second, we characterize six aggressive peer subgroups, including: peers using Tor, peers that bridge multiple p2p networks, and the top 10% of peers contributing to file availability. We find that these subgroups are more active in their trafficking, having more known CP and more uptime, than the average peer. Finally, while in theory Tor presents a challenge to investigators, we observe that in practice offenders use Tor inconsistently. Over 90% of regular Tor users send traffic from a non-Tor IP at least once after first using Tor.

June 2013

Digging up social structures from documents on the web, by Gessiou, E., Volanis, S., Athanasopoulos, E., Markatos, E.P., Ioannidis, S., 2012.

Abstract: We collected more than ten million Microsoft Office documents from public websites, analyzed the metadata stored in each document and extracted information related to social activities. Our analysis revealed the existence of exactly identified cliques of users that edit, revise and collaborate on industrial and military content. We also examined cliques in documents downloaded from Fortune-500 company websites. We constructed their graphs and measured their properties. The graphs contained many connected components and presented social properties. The a priori knowledge of a company's social graph may significantly assist an adversary to launch targeted attacks, such as targeted advertisements and phishing emails. Our study demonstrates the privacy risks associated with metadata by cross-correlating all members identified in a clique with users of Twitter. We show that it is possible to match authors collaborating in the creation of a document with Twitter accounts. To the best of our knowledge, this study is the first to identify individuals and create social cliques solely based on information derived from document metadata. Our study raises major concerns about the risks involved in privacy leakage due to document metadata.

Jan 2013

Distinct Sector hashing for Target Detection, by Joel Young, Kristina Foster, Simson Garfinkel, Kevin Fairbanks, December 2012.

Abstract: Using an alternative approach to traditional file hashing, digital forensic investigators can hash individually sampled subject drives on sector boundaries and then check these hashes against a prebuilt database, making it possible to process raw media without reference to the underlying file system.

Aug 2012

Infiltrate the Vault: Security Analysis and Decryption of Lion Full Disk Encryption, by Omar Choudary, Felix Grobert, Joachim Metz, August 2012.

Abstract: With the launch of Mac OS X 10.7 (Lion), Apple has introduced a volume encryption mechanism known as FileVault 2. Apple only disclosed marketing aspects of the closed-source software, e.g. its use of the AES-XTS tweakable encryption, but a publicly available security evaluation and detailed description was unavailable until now.. We have performed an extensive analysis of FileVault 2 and we have been able to find all the algorithms and parameters needed to successfully read an encrypted volume. This allows us to perform forensic investigations on encrypted volumes using our own tools. In this paper we present the architecture of FileVault 2, giving details of the key derivation, encryption process and metadata structures needed to perform the volume decryption. Besides the analysis of the system, we have also built a library that can mount a volume encrypted with FileVault 2. As a contribution to the research and forensic communities we have made this library open source. Additionally, we present an informal security evaluation of the system and comment on some of the design and implementation features. Among others we analyze the random number generator used to create the recovery password. We have also analyzed the entropy of each 512-byte block in the encrypted volume and discovered that part of the user data was left unencrypted.

Mar 2012

Effective Digital Forensics Research Is Investigator-Centric, by Robert J. Walls, Brian Neil Levine, Marc Liberatore, Clay Shields, August 2011.

Abstract: Many technical mechanisms across computer security for attribution, identification, and classification are neither sufficient nor necessary for forensically valid digital investigations; yet they are often claimed as useful or necessary. Similarly, when forensic research is evaluated using the viewpoints held by computer security venues, the challenges, constraints, and usefulness of the work is often misjudged. In this paper, we point out many key aspects of digital forensics with the goal of ensuring that research seeking to advance the discipline will have the highest possible adoption rate by practitioners. We enumerate general legal and practical constraints placed on forensic investigators that set the field apart. We point out the assumptions, often limited or incorrect, made about forensics in past work, and discuss how these assumptions limit the impact of contributions.

PinDr0p: using single-ended audio features to determine call provenance, by Vijay A. Balasubramaniyan, Aamir Poonawalla, Mustaque Ahamad, Michael T. Hunter, Patrick Traynor, 2010.

Abstract: The recent diversification of telephony infrastructure allows users to communicate through landlines, mobile phones and VoIP phones. However, call metadata such as Caller-ID is either not transferred or transferred without verification across these networks, allowing attackers to maliciously alter it. In this paper, we develop PinDr0p, a mechanism to assist users in determining call provenance - the source and the path taken by a call. Our techniques detect and mea- sure single-ended audio features to identify all of the applied voice codecs, calculate packet loss and noise profiles, while remaining agnostic to characteristics of the speaker's voice (as this may legitimately change when interacting with a large organization). In the absence of verifiable call metadata, these features in combina- tion with machine learning allow us to determine the traversal of a call through as many as three different providers (e.g., cellular, then VoIP, then PSTN and all combinations and subsets thereof) with 91.6% accuracy. Moreover, we show that once we identify and characterize the networks traversed, we can create detailed fin- gerprints for a call source. Using these fingerprints we show that we are able to distinguish between calls made using specific PSTN, cellular, Vonage, Skype and other hard and soft phones from locations across the world with over 90% accuracy. In so doing, we provide a first step in accurately determining the provenance of a call.

Jan 2012

Design of a Digital Forensics Evidence Reconstruction System for Complex and Obscure Fragmented File Carving, by Vrizlynn L.L. Thing, Tong-Wei Chua, Ming-Lee Cheong, 2011.

Dec 2011

Security Aspects of Piecewise Hashing in Computer Forensics, by Baier, H., Breitinger, F., May 2011.

Abstract: Although hash functions are a well-known method in computer science to map arbitrary large data to bit strings of a fixed length, their use in computer forensics is currently very limited. As of today, in a pre-step process hash values of files are generated and stored in a database, typically a cryptographic hash function like MD5 or SHA-1 is used. Later the investigator computes hash values of files, which he finds on a storage medium, and performs look ups in his database. This approach has several drawbacks, which have been sketched in the community, and some alternative approaches have been proposed. The most popular one is due to Jesse Kornblum, who transferred ideas from spam detection to computer forensics in order to identify similar files. However, his proposal lacks a thorough security analysis. It is therefore one aim of the paper at hand to present some possible attack vectors of an active adversary to bypass Kornblum's approach. Furthermore, we present a pseudo random number generator being both more efficient and more random compared to Kornblum's pseudo random number generator.

August 2011

Forensic Carving of Network Packets and Associated Data Structures, by Robert Beverly, Simson Garfinkel, Gregory Cardwell, August 2011.

Abstract: Using validated carving techniques, we show that popular operating systems (\eg Windows, Linux, and OSX) frequently have residual IP packets, Ethernet frames, and associated data structures present in system memory from long-terminated network traffic. Such information is useful for many forensic purposes including establishment of prior connection activity and services used; identification of other systems present on the system's LAN or WLAN; geolocation of the host computer system; and cross-drive analysis. We show that network structures can also be recovered from memory that is persisted onto a mass storage medium during the course of system swapping or hibernation. We present our network carving techniques, algorithms and tools, and validate these against both purpose-built memory images and a readily available forensic corpora. These techniques are valuable to both forensics tasks, particularly in analyzing mobile devices, and to cyber-security objectives such as malware analysis.

July 2011

Theory and practice of flash memory mobile forensics, by Salvatore Fiorillo, 2009

June 2011

Judges' Awareness, Understanding, and Application of Digital Evidence, by Gary Craig Kessler, 2010

July 2010

Sleuth Kit and Open Source Digital Forensics Conference

The slides from the first ever Sleuth Kit and Open Source Digital Forensics Conference are now available online:

March 2010

Internet geolocation: Evasion and counterevasion ACM Computing Surveys (CSUR), Volume 42, Issue 1 (December 2009)

Abstract: Internet geolocation technology aims to determine the physical (geographic) location of Internet users and devices. It is currently proposed or in use for a wide variety of purposes, including targeted marketing, restricting digital content sales to authorized jurisdictions, and security applications such as reducing credit card fraud. This raises questions about the veracity of claims of accurate and reliable geolocation. We provide a survey of Internet geolocation technologies with an emphasis on adversarial contexts; that is, we consider how this technology performs against a knowledgeable adversary whose goal is to evade geolocation. We do so by examining first the limitations of existing techniques, and then, from this base, determining how best to evade existing geolocation techniques. We also consider two further geolocation techniques which may be of use even against adversarial targets: (1) the extraction of client IP addresses using functionality introduced in the 1.5 Java API, and (2) the collection of round-trip times using HTTP refreshes. These techniques illustrate that the seemingly straightforward technique of evading geolocation by relaying traffic through a proxy server (or network of proxy servers) is not as straightforward as many end-users might expect. We give a demonstration of this for users of the popular Tor anonymizing network.

February 2010

The 'CSI Effect': Does It Really Exist?, by The Honorable Donald E. Shelton

Abstract: Crime and courtroom proceedings have long been fodder for film and television scriptwriters. In recent years, however, the media's use of the courtroom as a vehicle for drama has not only proliferated, it has changed focus. In apparent fascination with our criminal justice process, many of today's courtroom dramas are based on actual cases. Court TV offers live gavel-to-gavel coverage of trials over the Internet for \$5.95 a month. Now, that's "reality television"!

Reality and fiction have begun to blur with crime magazine television shows such as 48 Hours Mystery, American Justice, and even, on occasion, Dateline NBC. These programs portray actual cases, but only after extensively editing the content and incorporating narration for dramatic effect. Presenting one 35-year-old cold case, for example, 48 Hours Mystery filmed for months to capture all pretrial hearings as well as the 2-week trial; the program, however, was ultimately edited to a 1-hour episode that suggested the crime remained a "mystery" ... notwithstanding the jury's guilty verdict.

January 2010

Using Graphics Processors for Parallelizing Hash-based Data Carving, by Sylvain Collange, Marc Daumas, Yoginder S. Dandass, and David Defour

Abstract: The ability to detect fragments of deleted image files and to reconstruct these image files from all available fragments on disk is a key activity in the field of digital forensics. Although reconstruction of image files from the file fragments on disk can be accomplished by simply comparing the content of sectors on disk with the content of known files, this brute-force approach can be time consuming.

This paper presents results from research into the use of Graphics Processing Units (GPUs) in detecting specific image file byte patterns in disk clusters. Unique identifying pattern for each disk sector is compared against patterns in known images. A pattern match indicates the potential presence of an image and flags the disk sector for further in-depth examination to confirm the match. The GPU-based implementation outperforms the software implementation by a significant margin.

September 2009

Developments in Cisco IOS Forensics by Felix Lindner

Abstract: Cisco System's routers running Cisco IOS are still the prevalent routing platform on the Internet and corporate networks. Their huge population, architectural deficiencies and hugely diverse version distribution make them a valuable target that gains importance as common operating system platforms are closed down and secured. This paper takes the position that the currently used, well accepted practices for monitoring, debugging and post mortem crash analysis are insufficient to deal with the threat of compromised IOS devices. It sets forth a different method that reduces the requirement for constant logging, favoring on- demand in-depth analysis in case of suspicion or actual device crashes. The paper concludes by presenting the current state in the development of software supporting the proposed method and requesting feedback from the community on the software's future directions.

June 2009

Recovery of Damaged Compressed Files for Digital Forensic Purposes, by Bora Park Savoldi, A. Gubian, P. Jungheum Park Seok Hee Lee Sangjin Lee

Abstract: Nowadays compressed files are very widespread and can be considered, without any doubt, with regard to the Digital Forensic realm, an important and precious source of probatory data. This is especially true when in a digital investigation the examiner has to deal with corrupted compressed files, which have been gathered in the collection phase of the investigative process. Therefore, in the computer forensic field, data recovery technologies are very important for acquiring useful pieces of data which can become, in a court of low, digital evidence. This kind of technology is used not only by law enforcement, but also by the multitude of users in their daily activities, which justify the relevant presence of tools in the software market which are devoted to rescue data from damaged compressed files. However, state-of-the-art data recovery tools have many limitations with regard to the capability of recovering the original data, especially in the case of damaged compressed files. So far, such recovery tools have been based on a which controls the signature/header of the file and, thus, provides the offset to the raw compressed data block. As a result, they cannot recover the compressed files if the first part of the raw compressed data block, which pertains to the header, is damaged or the signature/header block is corrupted. Therefore, in order to deal with this issue, we have developed a new tool capable of rescuing damaged compressed files, according to the DEFLATE compression scheme, even though the header block is missing or corrupted. This represents a new interesting opportunity for the digital forensic discipline.

April 2009

A Framework for Automated Digital Forensic Reporting, by Lt. Paul Farrell, March 2009

Abstract: Forensic analysis is the science of finding, examining and analyzing evidence in support of law enforcement, regulatory compliance or information gathering. Today, almost all digital forensic analysis is done by humans, requiring dedicated training and consuming man-hours at a considerable rate. As storage sizes increase and digital forensics gain importance in investigations, the backlog of media requiring human analysis has increased as well. This thesis tests today's top-of-the-line commercial and open source forensic tools with the analysis of a purpose-built Windows XP computer system containing two users that engaged in email, chat and web browsing. It presents the results of a pilot user study of the PyFlag forensic tool. Finally, it presents a technique to use software to do a preliminary analysis on media and provide a human readable report to the examiner. The goal of the technique is to perform rapid triaging of media and allow the human examiner to better prioritize man hours towards media with high return on investment.

February 2009

The impact of full disk encryption on digital forensics, ACM SIGOPS Operating Systems Review archive, Volume 42, Issue 3 (April 2008), Pages 93-98

Abstract: The integration of strong encryption into operating systems is creating challenges for forensic examiners, potentially preventing us from recovering any digital evidence from a computer. Because strong encryption cannot be circumvented without a key or passphrase, forensic examiners may not be able to access data after a computer is shut down, and must decide whether to perform a live forensic acquisition. In addition, with encryption becoming integrated into the operating system, in some cases, virtualization is the most effective approach to performing a forensic examination of a system with FDE. This paper presents the evolution of full disk encryption (FDE) and its impact on digital forensics. Furthermore, by demonstrating how full disk encryption has been dealt with in past investigations, this paper provides forensics examiners with practical techniques for recovering evidence that would otherwise be inaccessible.

January 2009

Forensic Investigation of the Nintendo Wii: A First Glance, by Dr. Benjamin Turnbull

Abstract: The closed nature of the Wii makes it a challenging game console for forensic analysis. This article takes a first look at the platform, discussing the various places where forensically interesting information may be hidden. There's also a reference to an interesting article about how the ability of wimotes to move avatars from one system to another documents a woman's affair while her husband was serving in Iraq.

November 2008

Data Acquisition from Cell Phone using Logical Approach, by Keonwoo Kim, Dowon Hong, Kyoil Chung, and Jae-Cheol Ryou

Abstract: The article discusses three approaches for acquiring data from cell phones: physically removing the flash RAM chips and reading them directly; reading the data out using the JTAG interface, and running software on the cell phone to extract the files at a logical level. The authors have built a logical extraction system and are working on a system based on JTAG.

October 2008

Advanced JPEG carving, by Michael I. Cohen

Abstract: Michael I. Cohen presents a fully automated carver which can carve fragmented JPEGs using typical fragmentation patterns. (Cohen argues the the DFRWS carving challenges do not represent typical fragmentation patterns.)

Threats to Privacy in the Forensic Analysis of Database Systems by Patrick Stahlberg, Gerome Miklau, and Brian Neil Levine

Abstract: This paper looks at residual data left behind in databases after DELETE, UPDATE, and VACUUM operations. The authors show that residual data is a real issue in databases, and that it's pretty easy to modify a database so that no residual data is left behind. MySQL with MyISM tables has clean delete, but InnoDB does not. Very much worth reading.

August 2008

Lest We Remember: Cold Boot Attacks on Encryption Keys by J. Alex Halderman, Seth D. Schoen, Nadia Heninger, William Clarkson, William Paul, Joseph A. Calandrino, Ariel J. Feldman, Jacob Appelbaum, Edward W. Felten

Abstract: Increasingly memory analysis is of interest in forensic research - both because new malware only resides in memory, and because memory analysis is frequently the only way for analysts to get the keys that are used to protect cryptographic file systems. In this paper the authors show that cryptographic keys in memory are vulnerable to exploitation after the computer is turned off. The authors show that the contents of dynamic RAM are retained seconds, and sometimes minutes, after power is turned off. By chilling the memory the data can be retained as long as necessary. And while most laptops wipe their memory when they reboot, the authors show that the chilled memory can be moved from one laptop that wipes to another laptop that does not wipe. Finally, the authors show that it is possible to find the cryptographic keys in memory and correct random bit errors by using the AES key schedule as an error-correction code. The authors demonstrate an attack USB stick which reboots a computer protected with BitLocker, finds the cryptographic keys, and then allows access to the cleartext information on the disk.

July 2008

The Challenges of Using an Intrusion Detection System: Is It Worth the Effort?, by Rodrigo Werlinger, Kirstie Hawkey, Kasia Muldner, Pooya Jaferian and Konstantin Beznosov.

Abstract: In this article, the authors conducted interviews with 9 IT security practitioners who have worked with IDSs performed ethnographic observations within an organization that was deploying a new IDS. They found that security practitioners were heavily divided on the value of the IDS, and learned that the an IDS really only generates value if the network is well-understood before the IDS is deployed.

Mobile Phone Forensics Tool Testing: A Database Drive Approach, by Ibrahim M. Baggili, Richard Mislan, Marcus Rogers, 2007

Abstract: Given that phones are increasingly a primary source of forensic information in many cases, we need to be sure that the tools that are used for forensic analysis present data that is accurate and repeatable. Unfortunately they frequently aren't because of there are so many different kinds of phones on the market and the forensic tools lag far behind the market.

Detecting File Fragmentation Point Using Sequential Hypothesis Testing, by Anandabrata Pal, Taha Sencar, Nasir Memon, 2008

Abstract: This DFRWS 2008 article presents an improved approach for carving fragmented JPEGs using sequential hypothesis testing. According to the authors, "The technique begins with a header block identifying the start of a file and then attempts to validate via SHT each subsequent block following the header block. The fragmentation point is identified when SHT identifies a block as not belonging to the file. By utilizing this technique, we are able to correctly and efficiently recover JPEG images from the DFRWS 2006 and 2007 test sets even in the presence of tens of thousands of blocks and files fragmented into 3 or more parts. The bifragment gap carving technique enhanced with SHT allows us to improve the performance result of DFRWS 2006 challenge test-sets, although the technique cannot be used for DFRWS 2007. We then show how Parallel Unique Path enhanced with SHT is able to recover all fragmented JPEGs from DFRWS 2006 and all recoverable JPEGs from 2007 challenge test-sets. As far as we are aware, no other automated technique can recover multi-fragmented JPEGs from the DFRWS 2007 test set."

An Empirical Analysis of Disk Sector Hashes for Data Carving, by Yoginder Singh Dandass, Nathan Joseph Necaise, Sherry Reede Thomas, 2008

Abstract: Authors Dandass et. al analyzed 528 million sectors from 433,630 unique files. They computed the CRC32, CRC64, MD5 and SHA-1 of each sector. Not surprisingly, they find that the MD5 and SHA-1s of the sectors are different if the sectors are different. They find 94 CRC64 collisions and 30 million CRC32 collisions. The conclusion is that, if you are search for a single sector or building a database of single sector hashes, you are better off building a database of CRC64s because they are easier to store and dramatically faster to calculate than the traditional hash functions, and they are nearly as accurate.