Presentation is loading. Please wait.

Presentation is loading. Please wait.

Lorenzo Martignoni, Elizabeth Stinson, Matt Fredrikson, Somesh Jha, John Mitchell RAID 2008 1.

Similar presentations


Presentation on theme: "Lorenzo Martignoni, Elizabeth Stinson, Matt Fredrikson, Somesh Jha, John Mitchell RAID 2008 1."— Presentation transcript:

1 Lorenzo Martignoni, Elizabeth Stinson, Matt Fredrikson, Somesh Jha, John Mitchell RAID 2008 1

2 Bot-infected Computers Botnets are used to perform nefarious tasks, such as: keystroke logging, spyware installation, denial-of-service (DoS) attacks, hosting phishing web sites or command-and-control servers, spamming, click fraud, license key theft 2

3 Disadvantage of Traditional Method Even the most effective malware detectors fail to detect more than 30% of malware seen in the wild. Traditional malware detectors are based on syntactic signatures Malware producers can easily generate malware variants capable of evading existing signatures. Malware detectors have a finite set of syntactic signatures, but malicious programs have in infitely mutable syntax 3

4 Behavior-based malware detection Detect high-level actions that financially motivate malware development & distribution keystroke logging data leaking proxying program download and execute 4

5 Semantic gap between models and monitored events Monitor execution of the program using an emulator Lowest level events in behavior specifications are system calls Malicious behaviors are described as sequences of essential actions E.g. What we see NtDeviceIo... NtOpenFile NtCreateSe... NtMapView... is dierent from the essential actions we need to identify download a file and execute it Behaviors Low-level events 5

6 Solutions Complex & high-level behaviors are decomposed into multiple layers. The lowest layer represents system call invocations. Upper layers have a richer semantics. E.g. Hierarchy of events used to specify download_exec 6

7 Contributions A behavior-speciation language that can be used to describe novel, semantically meaningful behaviors. A detector that identifies when a process performs a specified high-level action, regardless of the process's source-code implementation of the action. Our evaluation demonstrates that our detector can distinguish malicious execution of high-level behaviors from benign. 7

8 Behavior Graphs A behavior graph is a directed graph of a form that is adapted from and extends AND/OR graphs. 8

9 Behavior Graphs Internal nodes represent events (with formal parameters) 9

10 Behavior Graphs Edges represent predicates on events arguments 10

11 Behavior Graphs ORed edges represent events of which at least one has to occur 11

12 Behavior Graphs ANDed edges represent events that all have to occur (but can occur in any order) 12

13 Behavior Graphs Annihilator and replicator nodes represent events that destroy and duplicate resources 13

14 Behavior Graphs Acceptor nodes represent actions taken by our system when behaviors are matched 14

15 Matching Malicious Behaviors 15 OS events are passed to the lowest layer

16 Construction of behavior graphs They developed our graphs manually and iteratively through domain knowledge and analysis of tens of gigabytes of execution traces, obtained from multiple runs of 1. around fifteen standard applications 2. over one hundred specially-crafted programs, 3. several malicious programs. 16

17 Architecture of the system Customized Qemu that instruments the guest code to monitor system call invocations, to perform taint analysis, and to track local user input. A behavior matcher that receives events in real-time and tries to match each behavior graph loaded. 17

18 Spec. of Malicious Behavior RI stands for Remotely Initiated Tainted refers to data received over the network 18

19 Result on Malicious bots Blank entries denote behaviors not matched because the bot did not implement them 19

20 Results on Benign Applications Under 2 scenarios: UI refers to an experiment in which user input tracking was not used, and “UI” to one which enabled User input tracking is very important to distinguish between behaviors triggered by the user and behaviors triggered automatically 20

21 Comments False Positive Problem!! E.g. Automatic Windows Update 21


Download ppt "Lorenzo Martignoni, Elizabeth Stinson, Matt Fredrikson, Somesh Jha, John Mitchell RAID 2008 1."

Similar presentations


Ads by Google