A FRAMEWORK FOR ANALYSIS AND COMPARISON OF DYNAMIC MALWARE ANALYSIS TOOLS

International Journal of Network Security & Its Applications (IJNSA), Vol.6, No.5, September 2014
A FRAMEWORK FOR ANALYSIS AND COMPARISON
OF DYNAMIC MALWARE ANALYSIS TOOLS
Waqas Aman
Norwegian Information Security Laboratory, Gjøvik University College, Norway
ABSTRACT
Malware writers have employed various obfuscation and polymorphism techniques to thwart static analysis
approaches and bypassing antivirus tools. Dynamic analysis techniques, however, have essentially
overcome these deceits by observing the actual behaviour of the code execution. In this regard, various
methods, techniques and tools have been proposed. However, because of the diverse concepts and
strategies used in the implementation of these methods and tools, security researchers and malware
analysts find it difficult to select the required optimum tool to investigate the behaviour of a malware and to
contain the associated risk for their study. Focusing on two dynamic analysis techniques: Function Call
monitoring and Information Flow Tracking, this paper presents a comparison framework for dynamic
malware analysis tools. The framework will assist the researchers and analysts to recognize the tool’s
implementation strategy, analysis approach, system-wide analysis support and its overall handling of
binaries, helping them to select a suitable and effective one for their study and analysis.
KEYWORDS
Malware Analysis, Functional Call Monitoring, Information Flow Tracking, System Call, Hooks.
1. INTRODUCTION
The internet has evolved exponentially since it was first launched in 1960. It has grown
increasingly from a four node communication model to a highly interconnected and sophisticated
network providing more and more services inheriting boundless benefits. The marketplace of
products and services are broadening on a global level, new ways of social interactions are
established and online banking has given new dimensions to banking operations. However, these
benefits have also motivated the people with malicious intents who are looking for any
opportunities to fulfill their objectives. Malware is one of the key mediums to furnish these
malicious intents.
Legitimate users are protected from such malicious codes by the security vendors who provide
them a software, antivirus, which identify and analyze these codes and alert the user accordingly.
Typically, antivirus is equipped with a signature database which is used in the matching process
to identify potential known or common threats. Malware analyst obtains a piece of suspected code
and analyzes it to find whether it is harmful or not. When a threat in a code is confirmed, the
analyst looks for a particular pattern of the threat and develops a signature for that code (malware)
and is added to the signature database to confront a particular malware in the future. This manual
process though seems trivial and does the job but is subjected to time consumption and errors as
there exist a number of variant of the same code. Statistics shows that anti-malware vendors are
experiencing thousands of malicious codes each day. McAfee, for instance, received more than
20,000 malware samples of mobile alone in 2012 [1]. According to NQ Mobile, mobile malwares
DOI : 10.5121/ijnsa.2014.6505 63

alone has grown up to 163% and has increased to more than 65000 in 2012 [2]. Hence there is a
need to automate the process to support quick, effective and timely analysis of such samples.
Typically, two methods are used to perform malware analysis; Static and Dynamic analysis. The
distinguishing ground between the two techniques is that the dynamic analysis observes a
malicious behavior while a sample code is executed whereas; the static approach does not execute
the code. Though, they differ in the techniques used, a number of methods and tools were
proposed to serve the same objectives of malware analysis. In this paper, two most common
techniques used in dynamic malware analysis: Function call monitoring and Information Flow
Tracking (IFT) will be conversed. Understanding these two techniques gives a general and
holistic picture of a malware executable. Functional Call Monitoring aims to investigate the
functional behavior and IFT provides and understanding of what data a malicious binary is
interested in during execution. Thus, provides the analyst a fair idea of: for what reasons does a
malware misbehave. However, analyzing a malware has turned into a crucial as well as an
essential skill for security professionals and forensics investigators. Malware analysis not only
enables an analyst to realize the malevolent code objectives but also gives an understanding of the
evolving trend of a malware thus providing an analyst a tool to enrich the detection techniques.
Due to the diversity of concept and strategies used in the implementation of these methods and
tools, security researchers and analyst finds it difficult to select an appropriate tool for their
investigations. Therefore, a framework is presented that aims to provide a platform to identify and
understand the techniques that various tools offer to an analyst and the strategies they implement
in accomplishing dynamic malware analysis thus helping them to select an optimum method or
tool for their investigations.
The rest of the article is organized as follow: In section 2, an understanding of malware analysis
will be highlighted, and preliminaries of Function Call monitoring and Information Flow
Tracking will be conversed. In section 3, the proposed framework is presented. A summary of
selected tools to be analysed and compared will be briefly described in section 4. Section 5, will
provide a comparative analysis of the discussed tools and finally, the paper will be concluded in
section 6.
64
2. MALWARE ANALYSIS
Malware is a piece of software or program that deliberately fulfill the malicious attempts of an
attacker. Malware comes in many forms with different objectives. The common terms used to
classify different types of malwares are Trojan, worm and virus, etc. [3]. Traditionally, they were
developed either for fun factors to show one's capabilities or for highlighting weaknesses within a
system. However, today, these motivations have gone to the highest level of treachery. Now we
can see a spectrum of motivation ranging from personal to national level interest and a whole new
underground economy is based on malwares these days [4]. Stuxnet [5] is one of the latest
pictures of such motivations. Malwares are propagated using numerous infection vectors such as
exploiting vulnerability on a client system, through an open or vulnerable network service, using
removable devices [5, 6] or through social engineering. To combat against malwares, systems are
now equipped with antivirus programs. Most of the antivirus programs consist of a scanner and a
signature database. The scanner matches file on the user system and matches them against the
available signatures. An alert is generated and the user is informed when a match is found.
Typically, two standard approaches are used to analyze the behavior of a malicious program.
Dynamic analysis is a set of methods that are used to understand the behavior of a program during
its execution while static analysis is used to investigate a program without executing it. The
following subsections will elaborate static analysis and the shortcomings associated with it.

65
2.1. Static Analysis
Analyzing a program to observe its behavior by investigating it without execution is commonly
known as static analysis. It can be performed in numerous ways depending upon the availability
of the code and its representation. Static analysis can assist in evaluating the memory errors and
can improve the correctness of a program execution if it source code is available [7, 8]. It can also
be used to inspect a binary executable with different tools [9]. Static analysis can be prompted
before or after dynamic analysis or can be done as a standalone procedure. Sometimes it is
performed to see if the analysts have missed anything suspicious after the dynamic analysis. And
is performed pre-dynamic analysis to analyze and understand the behavior prior to the code
execution in a live environment.
2.1.1. Shortcomings of Static Analysis
Because source code of most of the programs is not readily available makes the static analysis
approach harder to combat the malicious programs thus reduces its application. Analyzing
binaries with static approaches have inherited complications and challenges. Some of the
malwares with strong evasion and obfuscation techniques such as the presence of Opaque
Constants make the disassembling of a binary executable ambiguous due to which the resulting
code cannot be analyzed precisely [10]. Disassembling remains the essential part of the static
analysis of binaries which can be easily made ambiguous by simple obfuscation measures. Such
obfuscation techniques obscure the program flow, making the variables inaccessible and disable
the tracking of values stored in a register. These limitations of the static approach motivate the
development of analysis techniques that can overcome the mentioned code evasions and
transformations and to analyze a malicious program accurately and reliably.
2.2. Dynamic Malware Analysis
Dynamic analysis is the process of executing malware in a monitored environment to observe its
behaviors [11]. While a monitored program executes, it provide a detailed information readily
such as URL accessed, files created and accessed, information transferred, registry keys creation
and so forth. In this section, a detailed discussion will be made on two most common techniques
used in dynamic analysis: Information Flow Tracking and Function Call Monitoring.
Furthermore, a description of various approaches and concepts used to accomplish these
techniques will also be elaborated. The objective of the underlying discussion is to understand
two vital aspects of analyzing a malicious program i.e. their functional behavior and the data or
information they are interested in while they execute.
2.2.1. Information Flow tracking
Information Flow Tracking (IFT) approaches are used to monitor programs from the perspective
of how the program processes the data. During the analysis, suspected or interested data is tainted
or labeled before it is processed and its propagation is then observed when the data is processed.
Because of the tainting data, information flow tracking is also termed as Taint Analysis [12]. In
[13], the author illustrates information flow tracking process with a simple example reviewed
below in table 1. Considered a stack location, z, is an unsafe location. A received data, x, from the
network is considered suspicious and is tainted. During processing, x is copied to another
location, y. Now during program execution if the program jumps to the already tagged location z,
an alarm will be raised indicating a malicious activity.

66
Table1. Simple Information Flow Tracking
Program Execution Information Flow Tracking
….. Tag(z) = 1 Suspicious location is tagged unsafe
Received (&x); Tag(x) = 1 Unsafe data tagged
y = x; Tag(y) = Tag(x)
….. …..
Jmp z; Raise Alarm, as z is already tagged as unsafe & suspicious
Some of the key concepts used in the process of IFT analysis are described below:
• Tainted Sources and Destinations: Taint sources are the components of an Information
Flow Tracking system that labels interested or suspicious data in the system. They taint the
data and give them as an input to the system. Tainted destinations, on the other hand, are
the elements which respond to a tainted data introduced by the taint sources in a tailored
mechanism. Responses can either be triggering another source or raising an alert.
• Direct Tainted Data: As the name implies, direct tainted data refers to labelling an operand
directly. The common rule of data tainting is that when a tainted source value is moved or
copied to another operand, the destination operand is also tainted [14]. However, an IFT
system may contain more than one direct tainted operands involved in a single (or multiple)
instruction(s) with distinct labels as depicted in table 2. A guideline should be maintained
that direct how these tainted operand should be handled. For example, the most suspicious
or interested operand can be monitored over the other or they can be analyzed in a
combined manner under a common label.
Table 2. Direct Data Tainting
y = x; Tag(y) = Tag(x) Direct tainted data with distinct tags
y = y + x;
Tag(y) = Tag(x) + Tag(x) Direct tainted data with distinct
tags in an arithmetic operation
• Tainting Addresses: Pointer or address tainting is used when the information flow
tracking system is used to observe non-control data exploits or analyzing sensitive
information leakage and is done by generating an address using a tainted data [15]. When
an address A is tainted, it is detected as a malicious activity if A is de-referenced during
the program execution. Consider the example in table 3 below:
Table 3. Tainting Addresses
x = 0; Tag(y) = Tag(x) Direct tainted data with distinct tags
A = &x; Tainted Address A generated using tainted value X
A = 1; Tainted Address A is de-referenced – Generate Alarm
• Control Dependencies: Besides data and address tainting, information flow tracking can
also be performed using control flow tainting. For instance, an instruction X is control
dependent on instruction B, if B controls the execution of A [15]. An illustration of
control-flow tainting is provided in table 4. It implies that an address or data can only be
tracked if another tainted operand on which its execution is dependent is also processed
by the system.

67
Table 4. Control Flow Tracking
If (A == 0) { B == 1; } T ag(B) T ag(A) = 0
• Taint Policy: Taint policy is a set of rules that direct how a taint analysis should be
carried out within the information flow tracking system and covers three aspects of taint
analysis; taint Introduction, Propagation and Checking [15]. Taint introduction rules
direct how taint should be introduced to the system, propagation specifies how taints
statuses are to be updated for data derived from tainted or untainted operands and
checking is performed to validate the taint status and the corresponding action to be
taken.
• Over and Under-tainting: These are the two types of errors that can be experienced
during information flow tracking analysis [15]. Overtainiting occurs when the system
marks a data as tainted when it is not derived from a tainted source. Undertainting occurs
when the analysis process misses a value to be tainted in the information flow tainting.
2.3. Function Call Monitoring
A function is a subroutine or a code block in a program that execute specific operations. They are
used whenever a code or task is repeated. This property makes functions reusable and easy to
maintain if changes are required for a particular procedure. Functions are designed in an abstract
manner with more semantic oriented approach rather than focusing on the implementation details
which make it adaptable for many platforms. From the analysis perspective, such abstraction
provides an understanding of the overall behavior of a program. A function call is a simple
command which invokes a function by calling its name. Functions are usually monitored by
intercepting the calls made to it.
The process of hooking is used to capture the function calls. The monitored program is operated
in such a way that whenever a call to a function is made, the hook function is also invoked. Hook
is a function itself which implements the analysis procedure and performs tasks like logging
target program execution, observes intermediary function calls and analyzing various inputs and
outputs.
The get an understanding of dynamic analysis in the context of functions and function calls, the
under mention terminologies are explained to recognize how functions and function calls are
triggered, the tasks they perform and the level they execute on.
• Application Programming Interface: API is a set of programming specifications that
perform various functionalities. Operating Systems provide various API libraries of
object classes, functions, data structures, and variables that are grouped together to
accomplish common functions. They are offered at different layers of operating system.
Windows Cryptography API [16], for instance, focuses on the provision of constants,
command-line tools and functions, etc. required by the cryptography and certificate
services used in a Window operating system.
• System Calls: Operating Systems provides two modes of program execution: User and
Kernel mode. User programs such as word processing, imaging applications and
browsing, etc. are executed in the user mode while the operating system runs its own
programs in kernel mode. User mode processes don’t have a direct access to the system
level which can only be accessed by kernel mode process [17]. However, in order to
perform various tasks such as creating a file on disk, a user mode process have to call a

International Journal of Network Security Its Applications (IJNSA), Vol.6, No.5, September 2014
system level process. In operating systems, this is facilitated through special APIs called
System Calls. The user mode switches to the kernel mode when a system call is triggered
and the request can be executed then in a privileged mode at system level. Although there
are some kernel level malwares which gain privileges to extract sensitive information
such as Rootkits [18] however, most of the malwares infects user mode processes and
make their way to the privileged kernel mode. Another way of invoking system calls in
Windows operating systems are the Windows Native APIs. However, they differ from the
usual managed APIs (discussed earlier) which remain the same in every operating while
the native APIs varies from one service pack to another [19]. Some user mode processes
use native APIs as a communication medium between higher level APIs and the system
calls. Generally, a legitimate user mode application connects to the kernel level (system
calls) through Windows API however; a malware may bypass the managed API layer
triggering the native APIs directly to execute malicious code at the kernel level. As a
result, the dynamic analysis tools which use API hooking to observe malicious behavior
skip these malwares unnoticed.
• Hooking: API hooking enables the analyst to track specific behavioral elements of the
program execution such as intermediary calls and parameters used in a function. Hooking
to native interfaces can provide a truer and detail behavioral analysis of a malicious
function and provide an insight to reveal suspicious activities bypassing the managed
APIs. From the context of implementation, hooking can be performed in a variety of
ways to circumvent malicious code from its execution. The simple way of hooking onto a
function is to insert the hook function appropriately into the code, provided that the
source code is readily available. [20] provides a set of function in GNU Compiler
Collection (GCC) to support function hooking using flags during code compilation.
Binary analysis mechanisms, for instance, [18], can be used to detect and analyze
functional behavior, if binary executable are in hand.
Windows operating systems support function hooking using the Detours Library [21]. It
implements function hooking whenever a monitored function is called by diverting the
control flow to a hook. When the hook function is initialized for analysis, it then calls in
monitored function itself. Another way to achieve function hooking is debugging. It
allows the analyst to insert breakpoints into the code at defined functions to observe its
behavior. A number of techniques to debug Windows programs can be found at [22].
Breakpoints give a full control at the debugger level to access various process
components for analysis.
• The Trace: The resulting product of function hooking is termed as a function trace. It is
more like an analysis report containing rich information such as the parameters accessed
and processed and the extended functions called by the monitored function. However, a
raw trace file may be very complicated to understand and analyze. Different tools and
methods are available which sort the trace information in a required semantic
representation. Malicious and legitimate behavior of a function can be reviewed by using
a graph where the differences between the graphs can show the presence of malicious
activity [23].
68
3. PROPOSED FRAMEWORK
This section details the key concepts and strategies based on which the proposed framework is
developed. They are necessary to understand or analyse dynamic malware analysis tools.
3.1. Analysis Level and Environment
Analysis system can be implemented at different level of an operating system. Its design
assessment depends upon the type of malwares under investigation and the level of information

extraction. Kernel mode analysis can extract rich information such as the systems calls and the
intermediary Native APIs called. As a privileged mode operation, kernel mode analysis can hide
its execution from the user level malwares [18]. On the other hand, user mode analysis can
provide a detail description of the managed API calls and the invoked functions. However, it
cannot retrieve the information on the invoked native APIs and corresponding system calls.
Analyzing malwares in native environment can lead to infections therefore analysis is done in a
protected environment to analyze the monitored programs thoroughly without any risk of
infection. Analysis can be performed in an emulated environment which enables an analyst to
execute a binary program without being infected. The process is usually carried in a Sandbox
which allows a program to be executed in an environment where all systems components such as
I/O devices, CPU, memory and network services are emulated [24]. Similar kind of approach is
the Virtual Machine Monitoring (VMM) where analysis is carried out in a virtual, isolated
machine that runs on a real (host) system.
69
3.2. Monitored Process
A vital property of an analysis tool is its capacity to monitor to multiple processes at the same
time. Most of the analysis techniques observe a single process however; malwares usually tend to
be dependent on other actions (processes) or conditions to be triggered such as logic bombs [3].
Monitoring multiple processes enables an analyst to perform an in-depth investigation of
activities and execution paths executed by a malware. Thus, in order to evaluate an analysis tool
we can breakdown the analysis to single, spawned (where a monitored process activates a child
process) and system-wide process monitoring.
3.3. Analysis Method
Analysis techniques refer to the investigation methods which are used or available in a particular
tool. The two important analysis techniques focused this paper are the function call monitoring
and IFT. However, the implementation or analysis level of a tool restricts an analyst to investigate
a malware by using only few analysis methods. For instance, if an analysis tool is implemented in
user mode, an analyst cannot observe or monitor the system calls or the intermediary native APIs.
Thus as a thumb rule, tools incorporating more analysis methods and options provides a better
analysis environment and produce positive as well as detailed results.
3.4. Unpacking the Binaries
Packing is a process in which a binary executable is packed in a transformed version of a
program, which has a different syntax but have the same semantics. This technique is now used
by most of the malware writers to hide the malware semantics from an analysis tool using
different obfuscation and encryption measures. An unpacking module is appended with the
packed malware, which is activated in the memory of the victim’s system thus leaving no trace on
the secondary memory. Though, analysis can be performed after the malware is unpacked but it is
more advantageous for the analysts to perform analysis while it is packed as it enables them to
understand the unpacking techniques used. Various techniques such as Reconstruction [25], Write
xor Execute (W⊕X) Policy [26] or detecting packer with packer’ signature database is used to
handle packed binaries.
4. TOOLS AND TECHNIQUES
This section presents a summary of eight most cited tools and techniques that incorporate
Function Call monitoring and IFT to approach dynamic malware analysis. Furthermore, a brief

overview will be provided on how individual tools address the features and strategies discussed in
section 3.
Panorama [27] is a comprehensive malware analysis tool which implements data and address
tainting techniques of IFT. The given system provides a holistic system-wide investigation in an
emulated environment. An informationtrace in the form of a taint graph is presented during the
analysis of a monitored program and contains information like network packets payload, function
parameters and data accessed during the execution. While a program is emulated, the analysis is
performed at the kernel level to intercept the system calls and APIs are invoked. The system
designers, however, didn’t address the notion of unpacking the binaries in this tool.
CWSandbox [28] is a Function Call monitoring tool that supports both user and kernel mode
behavior analysis to observe calls made to the APIs and systems calls. A hook function, that
implements a monitoring function, is invoked each time when an API is called. A control process
is defined which monitors and report the analysis whenever a hook function is called. This
enables the analyst to monitor and analyze the child processes that can be launched by a malware.
To increase the performance of the analysis process, a custom DLL is compiled before the
analysis process in order to have a quick access to the required hook functions. API and systems
calls are written into a trace file that provides a top-bottom view of functions calls made during
the execution.
Ether [29] uses hardware virtualization techniques to monitor functional behavior of a program
on individual instruction level. A trap flag is setup when a monitored program is launched. This
enables the analyst to debug the program’s functional behavior as the trap flag is used as a
breakpoint during execution. The presented technique can only retrieve system call information
which is implemented using the SYSENTER instruction. The implemented flagging phenomenon
also enables tracking and recording of program instructions resulting in reporting of rich
information to the analyst detailing individual instruction execution.
TQana [30], is an Internet Explorer browser plug-in that analyze the dynamic behavior of a
Spyware. TQana can observe the functional behavior as well as the information trace of spyware
execution. The analysis performed is emulated at the kernel level where the monitoring program
has access to all the calls made. Tainting or IFT is implemented using the Navigate event of the
web browser which introduce taints to the system whenever a URL is entered in the address bar,
or a link is clicked on a web page. Furthermore, contents accessed by a web browser and
bookmarks are also tainted. API and System call monitoring is implemented using the Component
Object Model subsystem (COM).
For future attacks, malwares like Rootkits and Spywares hook themselves to system APIs and is
activated on desired events. This hooking is done in a stealthy mode to avoid detection.
Hookfinder [31] is a tool that is used to identify and detect these bogus hooks, presenting the
findings in a detailed report. Hookfinder is implemented in emulated environment where analysis
is carried out with kernel mode privileges. Hook detection is done by monitoring the control flow
of the system processing. The tool observes the data and address taints and whenever the control
flow (instruction pointer) is diverted to an unexpected API or system call (or a tainted value in
this case), it detects a hook and alerts the analyst. The analysis is done on a single-process level.
The analysis report presented reflects the specific API or function to which a hook was attached
and a trace of how the instruction pointer was manipulated.
Norman Sandbox [32] is a functional call monitoring system which is implemented in a simulated
environment where analysis is performed in the user space. It is designed to analyze worms and
viruses which use network shares and email as infection vectors. The authors argued that it can be
used to analyse packed malware binaries as the infection is already contained in a simulated
environment. However, they didn’t provide any unpacking technique instead suggested that in
70

order to increase analysis performance, packers can be detected with known databases containing
packer signatures. Moreover, it is advised to simulate all network services as some malware
intends to use various network ports and services, for instance, SMB 137/138/445 in case of
network sharing. If a read or write request to certain service is returned null, the malware might
terminate its existence resulting the analysis to fail entirely.
Justin [33] is a scheme to handle packed binaries. The theme of this model is that a malware can
be detected immediately after the binary is unpacked by its unpacking routine so that it is readily
available to be detected by any antimalware software. The scheme can be used as a plugin on the
top of an AV which doesn’t provide the analysis and detection of un-packer. The external anti-malware
software is activated once the Write xor Execute condition is detected as true in the
71
memory. The scheme is based on two assumptions:
i. Irrespective of the packing techniques used by a malware, the original binary will execute
from the same memory address after it is unpacked.
ii. After the unpacking, full control is given to the original packed binary.
Logically, these two assumptions remain valid for most of the malwares. However, they can be
seen as very strict rules and can be bypassed by malwares which evade detection techniques using
multiple layers of packing.
PolyUnpack [34] on the other hand is more focused on the un-packer execution. It observes the
dynamic behavior of the un-packer execution using both static and dynamic analysis. The packed
binary is disassembled and is analysed in a static approach while the packed binary is executed
and is observed dynamically at each instruction level. Both the operations are done at the memory
level. The packer code is identified by comparing the instruction trace in both the dissembled and
the executed binaries. If, in either of the analysis the trace is not matched, it indicates the packed
code is about to be unpacked.
5. COMPARATIVE ANALYSIS
This section presents an analysis of the summarized tools as per the proposed framework detailed
in section 3. The objective of this analysis is to provide the reader a clear idea of how the
mentioned tools and techniques incorporate implementation strategies and the support they
provide for function call monitoring and IFT analysis. A comparative analysis is shown in the
table 5.
It can be seen that most of the tools, Panorama, CWSandbox, TQana, HookFinder, Justin, have
focused the kernel level implementation to get a detail understanding of the malware. Malware
uses different APIs to call common system calls. They are more interested in system calls to
achieve privilege access. APIs provide a medium to connect to them which implies that analysis
concentration should be more focused on intercepting and analyzing system calls. This trend can
be seen in the studied tools where all of them provide mechanisms to analyze the system calls.
IFT is only offered in HookFinder, TQana and Panorama. However, the objective of performing
information tainting is different. HookFinder implements IFT to detect the stealthy hooks within a
system whereas, TQana and Panorama provide tainting techniques to monitor and analyze
sensitive information leakage.

72
Table 5. Comparative Analysis
Strategies Panorama
CW
Sandbox
Ether TQana
Hook
finder
Norman Justin
Poly
Unpack
Mode
User - - - -
Kernel - -
Environment
Emulated - - - - -
Virtual - - - - - -
Methods Support
API Hooking - - -
System Calls - -
Packer
Analysis
- - - - - -
Packer
Reconstruction
- - - - - - -
W ⊕ X Policy - - - - - - -
Trace - - - -
IFT - -
Process Scope
Single -
Spawned - - - - - - -
System Wide - - - - -
6. CONCLUSION
In this paper, two key techniques of dynamic analysis, IFT and Function Call Monitoring, were
detailed. Due to the diversity of concepts and strategies associated with these techniques, it is
hard to identify and select a suitable tool. A framework is presented which highlight the key
attributes of interest in dynamic malware analysis and provides a general yet detailed enough
reference to compare and analyse different analysis tools.
REFERENCES
[1] McAfee Labs. Mcafee threats report: Third quarter 2012, 2012. Online Available at:
http://www.mcafee.com/tw/resources/reports/rp-quarterly-threat-q3-2012.pdf . Last Accessed On:
10th July 2014
[2] NQ Mobile. Nq mobiles 2012 security report. Online Available at
http://en.nq.com/2012_NQ_Mobile_Security_Report.pdf. Last Accessed On: 10th July 2014
[3] A. Moser, C. Kruegel, and E. Kirda. Exploring multiple execution paths for malware analysis. In
Security and Privacy, 2007. SP ’07. IEEE Symposium on, pages 231–245, 2007.
[4] Jianwei Zhuge, Thorsten Holz, Chengyu Song, Jinpeng Guo, Xinhui Han, and Wei Zou. Studying
malicious websites and the underground economy on the chinese web. In Managing Information
Risk and the Economics of Security, pages 225–244. Springer US, 2009.
[5] David Kushner. The real story of stuxnet, Feb 2013. http://spectrum.ieee.org/telecom/security/the-real-
story-of-stuxnet. Last Accessed On: 10th July 2014.
[6] Hassen Saidi Phillip Porras and Vinod Yegneswaran. An analysis of conficker’s logic and
rendezvous points. Technical report, SRI International, 2009.
[7] Hao Chen, Drew Dean, and David Wagner. Model checking one million lines of c code. In In
Proceedings of the 11th Annual Network and Distributed System Security Symposium (NDSS,
pages 171–185, 2004.

[8] Henry Hanping Feng, Jonathon T. Giffin, Yong Huang, Somesh Jha, Wenke Lee, and Barton P.
Miller. Formalizing sensitivity in static analysis for intrusion detection. In In IEEE Symposium on
Security and Privacy, 2004.
[9] Mihai Christodorescu and Somesh Jha. Static analysis of executables to detect malicious patterns.
73
In In Proceedings of the 12th USENIX Security Symposium, pages 169–186, 2003.
[10] Andreas Moser, Christopher Kruegel, and Engin Kirda. Limits of static analysis for malware
detection, 2007.
[11] Blake Hartstein Matthew Richard Michael Ligh, Steven Adair. Malware Analyst’s Cookbook:
Tools and Techniques for Fighting Malicious Code. John Wiley Sons, 2010.
[12] William Enck, Peter Gilbert, Byung-Gon Chun, Landon P. Cox, Jaeyeon Jung, Patrick McDaniel,
and Anmol N. Sheth. Taintdroid: an informationflow tracking system for realtime privacy
monitoring on smartphones. In Proceedings of the 9th USENIX conference on Operating systems
design and implementation, OSDI’10, pages 1–6, Berkeley, CA, USA, 2010. USENIX
Association.
[13] Feng Qin, Cheng Wang, Zhenmin Li, Ho-seop Kim, Yuanyuan Zhou, and Youfeng Wu. Lift: A
low-overhead practical information flow tracking system for detecting security attacks. In
Proceedings of the 39th Annual IEEE/ACM International Symposium on Microarchitecture,
MICRO 39, pages 135–148, Washington, DC, USA, 2006. IEEE Computer Society.
[14] Asia Slowinska and Herbert Bos. Pointless tainting? Evaluating the practicality of pointer tainting,
2009.
[15] Edward J. Schwartz, Thanassis Avgerinos, and David Brumley. All you ever wanted to know
about dynamic taint analysis and forward symbolic execution (but might have been afraid to ask).
In In Proceedings of the IEEE Symposium on Security and Privacy, 2010.
[16] Cryptography reference: Microsoft Corporation, 2012. http://msdn.microsoft.com/en-us/
library/windows/desktop/aa380256(v=vs.85).aspx. Last Accessed On 10th July, 2014.
[17] User mode and kernel mode: Microsoft developer center-hardaware, 2012.
http://msdn.microsoft.com/en-us/library/windows/hardware/ff554836(v=vs.85).aspx. Last
Accessed On 10th July, 2014.
[18] Christopher Kruegel, William Robertson, and Giovanni Vigna. Detecting kernel-level rootkits
through binary analysis. In In Proceedings of the Annual Computer Security Applications
Conference (ACSAC), Tuscon, AZ, 2004.
[19] G Nebbett. Windows NT/2000 Native API Reference. New Riders Publishing, Thousand Oaks,
CA, SA, 2000.
[20] GCC–options of code generation conventions. http://gcc.gnu.org/onlinedocs/gcc-4.1.2/gcc/Code-
Gen-Options.html. Last Accessed On 10th July, 2014.
[21] Galen Hunt and Doug Brubacher. Detours: binary interception of win32 functions. In Proceedings
of the 3rd conference on USENIX Windows NT Symposium - Volume 3, WINSYM’99, pages
14–14, Berkeley, CA, USA, 1999. USENIX Association.
[22] Microsoft corporation: Debug hook function writing. http://msdn.microsoft.com/en-us/
library/z2zscsc2.aspx. Last Accessed On 10th July, 2014.
[23] Mihai Christodorescu, Somesh Jha, and Christopher Kruegel. Mining specifications of malicious
behavior. In In Proceedings of the 6th joint meeting of the European Software Engineering
Conference and the ACM SIGSOFT International Symposium on Foundations of Software
Engineering, pages 5–14, 2007.
[24] Fabrice Bellard. Qemu, a fast and portable dynamic translator. In Proceedings of the annual
conference on USENIX Annual Technical Conference, ATEC ’05, pages 41–41, Berkeley, CA,
USA, 2005. USENIX Association.

[25] Min Gyung Kang, Pongsin Poosankam, and Heng Yin. Renovo: a hidden code extractor for
packed executables. In Proceedings of the 2007 ACM workshop on Recurring malcode, WORM
’07, pages 46–53, New York, NY, USA, 2007. ACM.
[26] L. Martignoni, M. Christodorescu, and S. Jha. Omniunpack: Fast, generic, and safe unpacking of
malware. In Computer Security Applications Conference, 2007. ACSAC 2007. Twenty-Third
Annual, pages 431–441, 2007.
[27] Heng Yin, Dawn Song, Manuel Egele, Christopher Kruegel, and Engin Kirda. Panorama:
Capturing system-wide information flow for malware detection and analysis. In In Proceedings of
the 14th ACM Conferences on Computer and Communication Security (CCS07. DRAFT, 2007.
[28] C. Willems, T. Holz, and F. Freiling. Toward automated dynamic malware analysis using
74
cwsandbox. Security Privacy, IEEE, 5(2):32–39, 2007.
[29] Artem Dinaburg, Paul Royal, Monirul Sharif, and Wenke Lee. Ether: malware analysis via
hardware virtualization extensions. In Proceedings of the 15th ACM conference on Computer and
communications security, CCS ’08, pages 51–62, New York, NY, USA, 2008. ACM.
[30] Manuel Egele, Christopher Kruegel, Engin Kirda, and Heng Yin. Dynamic spyware analysis. In
Proceedings of the 2007 Usenix Annual Conference, 2007.
[31] Heng Yin, Zhenkai Liang, and Dawn Song. Hookfinder: Identifying and understanding malware
hooking behaviors. 2008.
[32] Norman Solutions. Norman Sandbox Whitepaper, 2003.
[33] Fanglu Guo, Peter Ferrie, and Tzi-Cker Chiueh. A study of the packer problem and its solutions. In
Proceedings of the 11th international symposium on Recent Advances in Intrusion Detection,
RAID ’08, pages 98–115, Berlin, Heidelberg, 2008. Springer-Verlag.
[34] P. Royal, M. Halpin, D. Dagon, R. Edmonds, and Wenke Lee. Polyunpack: Automating the
hidden-code extraction of unpack-executing malware. In Computer Security Applications
Conference, 2006, pages 289–300, 2006.
Waqas Aman received his bachelor’s degree in Computer Sciences from University of
Peshawar in 2008 and earned his MS-InfoSec degree from National University of Sciences
Technology Pakistan in 2011. He is currently pursuing his Ph.D. in Information Security
at Gjøvik University College Norway, working as a Research Fellow in the Adaptive
Security for Smart Internet of Things in eHealth (ASSET) Project financed by the Research
Council of Norway. He also have industrial expertise in technical information security as
well as in product management and has earned various international certifications
including CEH, CEI, ECSA, ENSA, EDRP and ITIL® and remained Certified Instructor for EC-Council™
Technical Certifications in South Asia.

A FRAMEWORK FOR ANALYSIS AND COMPARISON OF DYNAMIC MALWARE ANALYSIS TOOLS

More Related Content

What's hot (20)

Similar to A FRAMEWORK FOR ANALYSIS AND COMPARISON OF DYNAMIC MALWARE ANALYSIS TOOLS (20)

Recently uploaded (20)

A FRAMEWORK FOR ANALYSIS AND COMPARISON OF DYNAMIC MALWARE ANALYSIS TOOLS