Malware Detection in Android Applications

International Journal of Trend in Scientific Research and Development (IJTSRD)
Volume 3 Issue 5, August 2019 Available Online: www.ijtsrd.com e-ISSN: 2456 – 6470
@ IJTSRD | Unique Paper ID – IJTSRD26449 | Volume – 3 | Issue – 5 | July - August 2019 Page 2401
Malware Detection in Android Applications
Mr. Tushar Patil, Prof. Bharti Dhote
Department of Computer Engineering, SIT Lonawala, SPPU, Pune, India
How to cite this paper: Mr. Tushar Patil |
Prof. Bharti Dhote "Malware Detection in
Android Applications" Published in
International
Journal of Trend in
Scientific Research
and Development
(ijtsrd), ISSN: 2456-
6470, Volume-3 |
Issue-5, August
2019,pp.2401-2403,
https://doi.org/10.31142/ijtsrd26449
Copyright © 2019 by author(s) and
International Journal ofTrend inScientific
Research and Development Journal. This
is an Open Access article distributed
under the terms of
the Creative
Commons Attribution
License (CC BY 4.0)
(http://creativecommons.org/licenses/by
/4.0)
ABSTRACT
Android is a Linux-based operating system used for smart-phone devices.
Since 2008, Android devices gained huge market share due to its open
architecture and popularity. Increased popularity of the Android devices and
associated primary benefits attracted the malware developers. Rate of
Android malware applications increased between 2008 and 2016. In this
paper, we proposed dynamic malware detection approach for Android
applications. In dynamic analysis, system calls are recorded to calculate the
density of the system calls. For density calculation, we used two different
lengths of system calls that are 3-gram and 5-gram. Furthermore,NaiveBayes
algorithm is applied to classify applications as benign or malicious. The
proposed algorithm detects malware using 100 real-world samples of benign
and malware applications. We observe that proposed method gives effective
and accurate results. The 3-gram NaiveBayes algorithmdetects84%malware
application correctly and 14% benign application incorrectly. The 5-gram
Naive Bayes algorithm detects 88% malware application correctly and 10%
benign application incorrectly.
KEYWORDS: Malware Detection • Naive Bayes Classifier • System Calls •
Frequency • Density
INTRODUCTION
Android is a most popular and fastest growing mobile
application development framework. Since 2008, the
adoption rate of Android has increased quickly. There are
approximately 1.5 million Android devices being activated
every day[18]. In the first quarter of 2017, Android occupy
approximately 86.1% market share[17]. Itis anopen-source
platform based on Linux kernel. Android OS is developed
and maintained by Google and promoted by Open Handset
Alliance. Android applications aredevelopedin Java,Python.
Android provide very user-friendly functionalities at truly
low cost. Android users use Android phones for storage,
communication, the Internet surfing, etc. To analyze
malware static, dynamic and hybrid analysis methods are
used[1]. Static analysis method identifies malware by
unpacking and decompiling application.Mostly,Commercial
anti-virus uses signature-based malware detection
technique.Thedynamic analysis identifies malwarebehavior
after deploying and executing the application. The hybrid
analysis is the combination of static and dynamic methods.
There are two main steps to overcome malware named as
identification of malware and detection of malware.
Application signature, permissions, and Dalvik bytecodeare
the parameters used for static analysis of malware[3].
System calls, network traffic, user interactions are the
parameters used by dynamic analysis[3]. Hybrid analysis
technique uses a combined feature of static and dynamic
approach.
In this work, we describe dynamic malware detection
techniques. For dynamic analysis first, we install allsamples
on Emulator. Then, we run all applications for a 2-3 minute
and record system calls. After that, we calculate the
frequency of the system call. Next, we apply the filter on
system calls. Filtered system calls are used for calculating
density. Furthermore, system calls are parsed and mapped
into the machine learning algorithm. We use Naive Bayes
classifier for classification of the application as benign or
malicious. Using mapped system calls as input, we train the
classifier. After that, we apply classifier and classify the
application as benign or malicious. The whole system
applied to real world benign and malware application
samples.
Our main contribution in this work is: 1. we performed
system call based dynamic malware analysis techniques. We
used Naive Bayes classification algorithm for detection. 2.
We used 3-gram and 5-gram length of system calls which
reduces time complexityofsystem. While,wefilteredsystem
calls on basis of their frequency.It reducesoverheadwithout
losing accuracy. 3. Performance of the overall dynamic
malware detection system is better and gives more accurate
results. Proposed system gives 85% and 89% accuracy in
results for 3-gram and 5-gram Naive Bayes algorithm.
LITERATURE SURVEY
Faruki Parvez et al.[1] and Arshad Saba et al.[2], gives a
detailed survey of Androidarchitectureandmalware.Parvez
Faruki et al.[1], gives Android security architecture and its
issues, malware types, and its penetration techniques. They
discussed malware detection methods that are static
malware detection and dynamic malware detection. Also,
they covered malware analysis and detection approaches
according to their goal, methodology, and deployment.
Finally, they proposed a hybrid approach to analyze and
IJTSRD26449

International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
detect Android Malware. Arshad Saba et al.[2] gives details
of different Android malware types and its penetration
techniques. They categorized different antimalware
techniques like static and dynamic malwaredetection.Atthe
end, they proposed the hybrid antimalware concept to
overcome limitations of the static and dynamic approach.
Feizollah et al.[3] provide details about feature selection
from Android applications for malware detection. Based on
deep research, they categorized four different feature
selection group like application meta-data, hybrid, dynamic
and static features. It gives a novel introduction about
Android malware detection types and related features.They
proposed permission, signature, Java’s code, etc. features
used for static malware detection. While the system calls,
network traffic, user interactions are the feature set for
dynamic malware detection.
In paper [4], [5], [6], [7] authors suggested static malware
analysis techniques with a different approach. Geoffroy
Gueguen et al.[4] propose static malware analysis tool
named as Androguard. Androguard is the Python-based
static malware analysis tool used to disassemble and
decompile Android apps by using reverse engineering.
Androguard calculates application similarities and
differences by using NCD(Normalized Compression
Distance), fuzzy risk score and signatures of the malicious
application. Faruki Parvez et al.[5] describe the tool
Androsimilar. Androsimilar is a signature based static
malware analysis tool.Androsimilarautomatically generates
the signature of the test application. Generated signature is
compared againstmalwaresignaturedatabase.Thenidentify
it as the normal or malicious application. Daniel Arp et al.[6]
propose the static malware analysis tool called as Drebin.
Drebin is a static malware analysis tool which detects
malicious application directly on Android phone. Drebin
collects various features from application codeandmanifest
file. Then machine learning approach is used to distinguish
normal and malicious application. Sanz Borja et al.[7]
propose permission based static malware detection tool
called PUMA. PUMAsextractapplicationpermissionfrom the
manifest file. Then use the machine learning algorithm to
identify normal and malicious permissions
In paper [8],[9],[10],[11] authors suggest dynamic malware
analysis techniques with the different approach. Suarez
Tangil et al.[8] proposes the dynamic analysis tool named as
AlterDroid. AlterDroid a tool for dynamic analysis of hidden
malware distributed over application components.
Alterdroid analyses the behavioral difference between
original application and fault injected application. It creates
behavior signatures for both applications. It then analyze
differential signature with the helpofpatternmatching.Tam
Kimberly et al.[9], describe tool CopperDroid. CopperDroid
is virtual machine based automatic dynamic analysis system.
It reconstructs the behaviour of Android malware by
monitoring system calls. Shabtai Asaf et al.[10] suggest tool
Andromaly. Andromaly is the host-based malwaredetection
tool. Andromaly continuously monitors various metrics of
the device like battery usage, CPU usage, the number of
active processes and amount of data transferred through a
network. Then it applies the machine learning algorithm for
classifying data as normal and malicious. Lok Kwong Yan et
al.[15] proposes the dynamic malware tool called as
Droidscope. Droidscope is a dynamic malware analysis
platform which is based on virtual machine introspection.
Droidscope is built upon QEMU emulator. It is monitoring
whole operating system to get more information regarding
malware and also detect kernel level attack.
PROPOSED METHODOLOGY
Preprocessing: The first step of proposed systemis tocollect
real-world samples of benign and malware applications.
After collection of application sample, system next go to the
second step of recording system calls. Figure 1 shows the
flow of system call recording. Initially, we installed every
application on Android emulator and run for a 2-3 minute.
After that, we recorded system calls of each application and
copied into an external file(.csv). To trace system calls we
used.We know that each line in trainingsetrepresents single
application features with multiple feature integer and
feature values. Now we labeled each line that means each
application with 1 or 0. Where 1 means benign application,
and 0 means a malicious application in training set. We have
used 70% of application from system data set for training
data and remaining for testing. After all this data
preprocessing, we applied Naive Bayes classifierinnextstep
Algorithm 1: Naive Bayes Algorithm for Malware
Detection
 The duplicated files are mapped witha singlecopy ofthe
file data by mapping with the existing file data in the
cloud
 The comprehensive requirements in multi-user cloud
storage systems and introduced the model of
deduplicatable dynamic PoS.
Input: Android Application System calls stored in .csv file
Output: Class from which given system calls belong.
1. Foreach line in file .csv do
2. Remove all parameters except system call name;
3. Store all system call names in another file called system
call name;
4. End
5. Foreach system call name in file system call name do
6. Assign unique integer number;
7. Store all integers in file integer system call file;
8. End
9. Foreach integer system call do
10. Calculate 3-gram and 5-gram length;
11. End
12. Foreach length of system call
13. Compute frequency of each integer then;
14. Foreach system call if frequency is less than 100;
15. Remove from file ;
16. Compute density of each integer then;
17. Store data into value pair format in data file;
18. End
19. After all this data processing apply Naive Bayes
classifier.
20. Foreach class instance
21. Calculate prior probability;
22. P(C) = Nc N
23. End
24. Foreach known value pair
25. Calculate conditional probability;
26. P(w|c) = countw, c() + 1/count(c) + |V
27. | End
28. Foreach unknown value pair
29. Calculate posterior probability;

International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
30. Cmap = argmaxP(x1, x2, x3, , xn)P P(C) End
31. Compare posterior probabilityforeachclassthenreturn
class with highest probability as result.
Algorithms
Let D be the Whole system which consists,
D= {I, P, O}
Where,
Q- Users Query {q1, q2…, qN}
P- Procedure,
F-Files set of {f1, f2,…,fn}
I-Input,
I-{F, Q},
O- Output.
Where:
F = Represents the file,
m1, m2, m3, m4= representing the ith block of the file,
e = encryption key
Phase 1: Pre-process Phase
In the pre-processing phase,
e← H(F), id ← H(e).
Then, the user announces that it has a certain file via id. If
the file does not exist, the user goes into the upload phase.
Otherwise, the user goes into the De-Duplication phase.
Phase 2 The Upload File
(C, T )← Encoding(e, F)
Let the file F = (m1, . . . ,mn).
The user first invokes the encoding according
Phase 3. The De-Duplication Data(file)
res∈ {0, 1} ← De-Duplication {U(e, F), S(T)}
If a file announced by a user in the pre-process phase exists
in the cloud server, the user goes into the De-Duplication
phase and runs the De-Duplication protocol
Phase 4: The Update File
res∈ {he∗, (C∗, T ∗)i,⊥} ← Updating{U(e, i, m, OP), S(C, T )}
In this phase, a user can arbitrarily update the file by
invoking the update protocol
Phase 5: The Proof of Storage to Owner
res∈ {0, 1} ←Checking{S(C, T ), U(e)}
At any time, users can go into the proof of storage phase if
they have the ownerships of the files. The users and the
cloud server run the checking protocol.
RESULT AND DISCUSSIONS
User can upload, download update on cloud server and
provide data De-Duplication.
CONCLUSIONS
In this work, we developed dynamic malware detection
system to detect malware in Android applications. For
dynamic detection, we used system calls invoked by the
application during execution. After that, Naive Bayes
classifier is used to classify runtime behaviorofapplications.
In addition, we used 3-gram and 5-gram length of system
calls. Instead of using every system calls; we filter system
calls based on frequency. Filtered system calls are used to
calculate density. Then, by applying Naive Bayes classifier,
we classified application in two different classes that are
benign and malware. For all system implementations, we
used real-world malware and benign application samples.
Proposed method gives more accurate results and performs
better than previous work. For 3-gram Naive Bayes
classifier, the system gives 85% accuracy while in 5-gram
Naive Bayes classifier; the system gives 89% accuracy. This
indicates the performance of the system is proportional to
the length of system calls.
REFERENCES
[1] Faruki Parvez, Ammar Bharmal, Vijay Laxmi, Vijay
Ganmoor, Manoj Singh Gaur, Mauro Conti, and
Muttukrishnan Rajarajan. ”Android security: a survey
of issues, malware penetration, and defenses.” IEEE
communications surveys & tutorials 17, no. 2(2015):
998-1022.
[2] Arshad Saba, Munam AliShah,AbidKhan,andMansoor
Ahmed. ”Android malware detection & protection: a
survey.” Int. J. Adv. Comput. Sci. Appl 7, no. 2 (2016):
463-475.
[3] Feizollah Ali, Nor Badrul Anuar, Rosli Salleh, and
Ainuddin Wahid Abdul Wahab. ”A review on feature
selection in mobile malware detection.” Digital
Investigation 13 (2015): 22-37.
[4] Desnos Anthony. ”Androguard: Reverse engineering,
malware and goodware analysis of android
applications.” URL code. google. com/p/androguard
(2013).
[5] Faruki Parvez, Vijay Ganmoor, Vijay Laxmi, Manoj
Singh Gaur, and Ammar Bharmal. ”AndroSimilar:
robust statistical feature signature for Android
malware detection.” In Proceedings of the 6th
International Conference on Security of Information
and Networks, pp. 152-159. ACM, 2013.
[6] Arp Daniel, Michael Spreitzenbarth, Malte Hubner,
Hugo Gascon, Konrad Rieck, and C. E. R. T. Siemens.
”DREBIN: Effective and Explainable Detection of
Android Malware in Your Pocket.” In NDSS. 2014.
[7] Sanz Borja, IgorSantos, Carlos Laorden, XabierUgarte-
Pedrero, Pablo Garcia Bringas, and Gonzalo lvarez.
”Puma: Permission usage to detect malware in
android.” In International Joint Conference CISIS12-
ICEUTE 12-SOCO 12 Special Sessions, pp. 289-298.
Springer Berlin Heidelberg, 2013.
[8] Suarez-Tangil, Guillermo, Juan E. Tapiador, Flavio
Lombardi, and Roberto Di Pietro. ”ALTERDROID:
differential fault analysis of obfuscated smartphone
malware.” IEEE Transactions on Mobile Computing15,
no. 4 (2016): 789-802.
[9] Tam Kimberly, Salahuddin J. Khan,AristideFattori,and
Lorenzo Cavallaro. ”CopperDroid: Automatic
Reconstruction of Android Malware Behaviors.” In
NDSS. 2015.
[10] ShabtaiAsaf,UriKanonov,YuvalElovici,ChananGlezer,
and Yael Weiss. ”Andromaly: a behavioral malware
detection framework for android devices.” Journal of
Intelligent Information Systems 38, no. 1 (2012): 161-
190.
[11] Yan, Lok-Kwong, and Heng Yin. ”DroidScope:
Seamlessly Reconstructing the OS and Dalvik Semantic
Views for Dynamic Android Malware Analysis.” In
USENIX security symposium, pp. 569-584. 2012.

Malware Detection in Android Applications

More Related Content

What's hot (20)

Similar to Malware Detection in Android Applications (20)

More from ijtsrd (20)

Recently uploaded (20)

Malware Detection in Android Applications