master's defense

38
Visualization techniques in Attack Graphs By: Ashok R Varikuti 05/18/2009 1

Upload: ashok-varikuti

Post on 28-Jan-2015

112 views

Category:

Technology


2 download

DESCRIPTION

My master's defense which discusses about various visualization techniques in attack graphs. I primarily used prefuse to generate a dynamic and interactive display for attack graphs.

TRANSCRIPT

Page 1: Master's Defense

Visualization techniques in Attack Graphs

By: Ashok R Varikuti

05/18/2009

1

Page 2: Master's Defense

Quote

“A picture is worth a thousand words.” – Napoleon Bonaparte

2

Page 3: Master's Defense

Outline

• Introduction to Attack Graphs

• Current limitations in Attack Graphs

• Survey of visualization toolkits

• Layered architecture of the model

• Discussion on how the model achieves the desired goal

• Implementation

• Conclusion3

Page 4: Master's Defense

Attack Graphs

• Provides a visual representation of attack paths and potential vulnerabilities in the network

• Makes the life of a system administrator easier by providing a convenient interface to identify vulnerabilities in the network

• MuLVAL generates attack graphs using a logic based approach

4

Page 5: Master's Defense

Attack Graphs

• Graph is generated using logic-based approach, hence Logical Attack Graph

• A node in the graph is a logical statement

• The edges in the graph specify the causality relations between network configurations and an attacker’s potential privileges.

• Illustrates “why the attack can happen”.

5

Page 6: Master's Defense

Logical Attack Graph Generator

6

Page 7: Master's Defense

3host network 7

Page 8: Master's Defense

Tree representation of 3host scenario attack graph8

Page 9: Master's Defense

Description

• The root node is the attack goal meaning “the attacker can execute arbitrary code as user root on machine workStation”

• The edges in the graph represent the “depends on” relation

• Fact Node 2 is enabled either by r2a or r2b, which are the derivation nodes for 2

9

Page 10: Master's Defense

10

Example:Energy Management Network

10

Page 11: Master's Defense

EMN

• EMN has 3 subnets, a DMZ (Demilitarized

Zone), an internal subnet, and an EMS(Energy Management System)

• Host-grouping applied based on similar configurations Ex: workstation

• The web server and the VPN server are directly accessible from the Internet

• The web server can access the file server through the NFS file-sharing protocol

11

Page 12: Master's Defense

EMN Continued..

• VPN server is allowed access to all hosts in the internal subnet

• Access to the EMS subnet is only allowed from the Citrix server in the internal subnet, and even then only to the data historian

• The attacker’s goal is to gain privileges to execute code on the communication Server

12

Page 13: Master's Defense

13

MulVAL LogicalAttackGraph

13

Page 14: Master's Defense

14

Limitations in Attack Graphs

• Attack Graph Problems: Size & Complexity– Difficult to quickly identify most important data – Difficult to assess and act on complete set of

possible attack steps– Not user interactive– Layout of the graph doesn’t correspond to the

underlying topology of the network

14

Page 15: Master's Defense

Contributions

• Solutions– Make the attack graph user interactive– Provide user interactivity, zooming and action

listener features – Provide different views of the graph for

analyzing in different dimensions

15

Page 16: Master's Defense

Visualization toolkits

• JUNG – An open source java software library. Mainly used in performing data analysis on relational data sets

• Piccolo – Mainly 2D graphics library with tree and fisheye layouts as salient features

• Graphviz – Open source package generally used to produce static visualizations

16

Page 17: Master's Defense

Two layered architecture 17

Page 18: Master's Defense

Static Layer

• Parse the text file into rich Dot Format

• The dot file is easy to analyze and produces a static image file as output.

• Disadvantages:– Layout

• Produces static images with random layout

– Scalability• Difficult to analyze large network's.• Possibility of edge-subnet, edge-edge and node-

node overlapping 18

Page 19: Master's Defense

Data Format

• Node (V):- The node is the most basic data type. It maps to an actual node in the real world enterprise network

• Gateway (G):- A gateway G typically maps to a firewall in a real world enterprise network.

• Subnet(S):- A subnet S is defined as :-

19

Page 20: Master's Defense

Data Format

• Connection (C):- A connection C is defined as the link between <subnet, gateway>.

• AttackStep (AS):- An attack step AS is defined as the link between <node,node>.

20

Page 21: Master's Defense

21

Topology-MappedAttack Graph

Improving Attack Graph Visualization through Data Reduction and Attack Grouping. John Homer, et al. In 5th International Workshop on Visualization for Cyber Security.

21

Page 22: Master's Defense

22

UntrimmedAttack Graph

Improving Attack Graph Visualization through Data Reduction and Attack Grouping. John Homer, et al. In 5th International Workshop on Visualization for Cyber Security.

22

Page 23: Master's Defense

23

TrimmedAttack Graph

Improving Attack Graph Visualization through Data Reduction and Attack Grouping. John Homer, et al. In 5th International Workshop on Visualization for Cyber Security.

23

Page 24: Master's Defense

Dynamic Layer

• Parse the Dot file using a dot grammar

• Prefuse conversion:-– Import prefuse packages– Prefuse graph object construction

• Map the parsed dot attributes with the prefuse attributes

• Construct the graph object based on the attribute values

24

Page 25: Master's Defense

The prefuse toolkit

• A java user interface toolkit for constructing interactive information visualization applications

• Supports visualization, animation, and interaction• Application building by stringing together fine-

grained, reusable components• Layers of indirection between source data,

visualized data and rendering.

25

Page 26: Master's Defense

System architecture

26

Page 27: Master's Defense

Toolkit features

• Data structures and I/O libraries• Multiple visualizations, multiple views • Application design through composable modules• A library of provided layout and distortion techniques• Animation and time-based processing• Graphics transforms, including panning and zooming• A full force simulator for physics-based interfaces• Interactor components for common interactions• Integrated color maps and search functionality• Event logging to support visualization evaluation

27

Page 28: Master's Defense

Architecture

• Filtering is the process of mapping abstract data to a representation suitable for visualization

• Actions are used to select visualized data and set visual properties such as location, font. Also used to perform tasks such as filtering, layout and color assignment

• Actionlists are used to enforce ordered execution of actions

28

Page 29: Master's Defense

Architecture Cont..

• Renderer's draw the visual items on to the screen by rendering the visual attributes

• The Display component draws all the visible items using appropriate renderer's onto the screen

• Display provides support to mouse and keyboard events on visible items

29

Page 30: Master's Defense

Useful features

• Panning and zooming allows the user to concentrate on the essential parts of the graph

• Expression techniques allows the user to visualize specific areas of graph with particular attributes Ex: The user can filter the graph to display information specific to the top three vulnerabilities

• Interactor components for common interactions such as showing the predicates attached with an edge

30

Page 31: Master's Defense

Extendible features

• Search functionality to search for a particular edge/node in the graph with a specific property.

• Overview feature allows to capture the whole view of network in a small dialog box

31

Page 32: Master's Defense

Implementation

32

Page 33: Master's Defense

Impl Contd..

33

Page 34: Master's Defense

34

Future Work

Page 35: Master's Defense

Two layered architecture 35

Page 36: Master's Defense

Conclusion

• The architecture generates an interactive display of the attack graphs.

• This provides a convenient way for the researcher's to enhance the tool in a better way.

• Provides a convenient way for network administrator to map the attack graph's into real network topology.

36

Page 37: Master's Defense

References

• http://prefuse.org• John Homer, Ashok Varikuti, Xinming Ou, and Miles A.

McQueen. Improving attack graph visualization through data reduction and attack grouping. In 5th International Workshop onVisualization for Cyber Security (VizSEC 2008), Cambridge, MA, U.S.A., September 2008.

37

Page 38: Master's Defense

Demo of Energy Management Network

38