a system call analysis method with mapreduce for malware detection 2011 ieee 17th international...

22
A System Call Analysis Method with MapReduce for Malware Detection 2011 IEEE 17th International Conference on Parallel and Distributed Systems Shun-Te Liu *, Hui-ching Huang* Information & Communication Security Lab TL, Chunghwa Telecom Co., Ltd. Yi-Ming Chen Department of Information Management National Central University 102062602 黃黃黃 1 /22

Upload: henry-phillips

Post on 28-Dec-2015

231 views

Category:

Documents


3 download

TRANSCRIPT

Page 1: A System Call Analysis Method with MapReduce for Malware Detection 2011 IEEE 17th International Conference on Parallel and Distributed Systems Shun-Te

1

A System Call Analysis Method with MapReduce for Malware Detection

2011 IEEE 17th International Conference on Parallel and Distributed Systems

Shun-Te Liu *, Hui-ching Huang*

Information & Communication Security Lab TL, Chunghwa Telecom Co., Ltd.

Yi-Ming Chen

Department of Information ManagementNational Central University

102062602

黃建忠

/22

Page 2: A System Call Analysis Method with MapReduce for Malware Detection 2011 IEEE 17th International Conference on Parallel and Distributed Systems Shun-Te

2

outline

Introduction Detect malware behavior Evaluation Conclution

/22

Page 3: A System Call Analysis Method with MapReduce for Malware Detection 2011 IEEE 17th International Conference on Parallel and Distributed Systems Shun-Te

3

Malware by categories

/22

Page 4: A System Call Analysis Method with MapReduce for Malware Detection 2011 IEEE 17th International Conference on Parallel and Distributed Systems Shun-Te

4

How to detect malware

Signature-based approach

Behavior-based approach

/22

Page 5: A System Call Analysis Method with MapReduce for Malware Detection 2011 IEEE 17th International Conference on Parallel and Distributed Systems Shun-Te

/225

Behavior-based approach

Detect malware by real-time monitoring mechanisms

Ex: system call monitoring (procMon)

Page 6: A System Call Analysis Method with MapReduce for Malware Detection 2011 IEEE 17th International Conference on Parallel and Distributed Systems Shun-Te

6

Malicious behavior patterns

Privacy invasion

Self-replication

Persistent behavior

/22

Page 7: A System Call Analysis Method with MapReduce for Malware Detection 2011 IEEE 17th International Conference on Parallel and Distributed Systems Shun-Te

7

Mordern malware

Discrete behavior download malicious module

Module-base malware driver or DLL

/22

Page 8: A System Call Analysis Method with MapReduce for Malware Detection 2011 IEEE 17th International Conference on Parallel and Distributed Systems Shun-Te

/228

requirements

the collected and analyzed data is much richer (system calls)

module dependency

Page 9: A System Call Analysis Method with MapReduce for Malware Detection 2011 IEEE 17th International Conference on Parallel and Distributed Systems Shun-Te

/229

Client–server model

Page 10: A System Call Analysis Method with MapReduce for Malware Detection 2011 IEEE 17th International Conference on Parallel and Distributed Systems Shun-Te

10

MapReduce

A programming model for processing large data sets with a parallel, distributed algorithm on a cluster

Apache Hadoop

/22

Page 11: A System Call Analysis Method with MapReduce for Malware Detection 2011 IEEE 17th International Conference on Parallel and Distributed Systems Shun-Te

11

Persistent behavior

Malware

ASEP ( auto-start extensibility point)

Remain alive after system reboot

/22

Page 12: A System Call Analysis Method with MapReduce for Malware Detection 2011 IEEE 17th International Conference on Parallel and Distributed Systems Shun-Te

12

ASEP(1)

Can be a file or registry keys

Ex: autorun.ini

/22

Page 13: A System Call Analysis Method with MapReduce for Malware Detection 2011 IEEE 17th International Conference on Parallel and Distributed Systems Shun-Te

/2213

ASEP(2) HKLM\SOFTWARE\Microsoft\ Windows\CurrentVersion\Run

HKLM\SOFTWARE\Microsoft\ Windows NT\ CurrentVersion\

Winlogon\Notify (dll)

HKLM\System\CurrentControlSet\ (driver)

Page 14: A System Call Analysis Method with MapReduce for Malware Detection 2011 IEEE 17th International Conference on Parallel and Distributed Systems Shun-Te

/2214

Persistent behavior module(1)

Page 15: A System Call Analysis Method with MapReduce for Malware Detection 2011 IEEE 17th International Conference on Parallel and Distributed Systems Shun-Te

15

Persistent behavior module(2)

/22

Page 16: A System Call Analysis Method with MapReduce for Malware Detection 2011 IEEE 17th International Conference on Parallel and Distributed Systems Shun-Te

16

Dependency Relationship(1)

ASEP is seen as a part of module white list filter

/22

Page 17: A System Call Analysis Method with MapReduce for Malware Detection 2011 IEEE 17th International Conference on Parallel and Distributed Systems Shun-Te

17

Dependency Relationship(2)

Mi Mj

/22

Page 18: A System Call Analysis Method with MapReduce for Malware Detection 2011 IEEE 17th International Conference on Parallel and Distributed Systems Shun-Te

18

Dependency structure matrix

Check diagonal cells

A B , B C , C A

/22

Page 19: A System Call Analysis Method with MapReduce for Malware Detection 2011 IEEE 17th International Conference on Parallel and Distributed Systems Shun-Te

19

Accuracy

/22

Page 20: A System Call Analysis Method with MapReduce for Malware Detection 2011 IEEE 17th International Conference on Parallel and Distributed Systems Shun-Te

20

Performance

/22

Page 21: A System Call Analysis Method with MapReduce for Malware Detection 2011 IEEE 17th International Conference on Parallel and Distributed Systems Shun-Te

21

contribution

Propose a relation-based method to correlate the discrete behavior of malware.

Implement a prototype of Maltrix on the Hadoop platform.

/22

Page 22: A System Call Analysis Method with MapReduce for Malware Detection 2011 IEEE 17th International Conference on Parallel and Distributed Systems Shun-Te

22

challenges

Some malwares don’t require ASEP

The cost of data transmission hasn't been measured.

Anti-api hooking

Without using system calls

/22