Method of modeling behavior pattern of instruction set in N-gram manner, computing device operating with the method, and program stored in storage medium to execute the method in computing device

• US 10,007,788 B2
• Filed: 02/05/2016
• Issued: 06/26/2018
• Est. Priority Date: 02/11/2015
• Status: Active Grant

• Current Assignee: Electronics and Telecommunications Research Institute
• Original Assignee: Electronics and Telecommunications Research Institute
• Inventors: Moon, Dae Sung, Kim, Ik Kyun, Lee, Han Sung
• Primary Examiner(s): Cribbs, Malcolm
• Application Number:
  US15/017,504
• Publication Number:
  US 20160232345A1
• Time in Patent Office: 872 Days
• Field of Search: None
• CPC Class Codes: G06F 21/566 Dynamic detection, i.e. detection performed at run-time, e.g. emulation, suspicious activities

Abstract

A computing device configured to execute an instruction set is provided. The computing device includes a system call hooker for hooking system calls that occur by the instruction set while the instruction set is executed, a category extractor for extracting a category to which each of the hooked system calls belongs with reference to category information associated with a correspondence relationship between a system call and a category, a sequence extractor for extracting one or more behavior sequences expressed in an N-gram manner from a full sequence of the hooked system calls with reference to the extracted category, and a model generator for generating a behavior pattern model of the system calls that occur when the instruction set is executed, based on a number of times that each of the extracted behavior sequences occurs.

Claims

1. A non-transitory computer-readable storage medium storing a program configured to model a behavior pattern associated with system calls that occur by an instruction set executed in a computing device, the program executing a process, in the computing device, that comprises:

hooking, by a processor of the computing device, the system calls while the instruction set is executed under a control of the processor;

extracting, by the processor, a category to which each of the hooked system calls belongs, with reference to category information stored in at least one of a first storage of the computing device or a second storage provided separately from the computing device;

extracting, by the processor, one or more behavior sequences expressed in an N-gram manner from a full sequence of the hooked system calls, with reference to the extracted category;

generating, by the processor, a model of the behavior pattern based on a number of times that each of the extracted N-gram behavior sequences occurs;

comparing at least one of the generated model of the behavior pattern or the stored model of the behavior pattern with a reference model; and

determining, based on the comparison, whether the executed instruction set is malicious or normal.

2. The storage medium of claim 1, wherein the model of the behavior pattern is expressed in a vector format, and the vector format is generated by associating all behavior sequences expressed in the N-gram manner with the number of times that each of the extracted behavior sequences occurs.

3. The storage medium of claim 1, wherein the process further comprises storing information associated with the model of the behavior pattern in at least one of the first storage or the second storage.

4. A method for modeling a behavior pattern associated with system calls that occur by an instruction set executed in a computing device, the method comprising:

hooking, by the computing device, the system calls;

extracting, by the computing device, a category to which each of the hooked system calls belongs;

extracting, by the computing device, one or more behavior sequences expressed in an N-gram manner from a full sequence of the hooked system calls, with reference to the extracted category;

generating, by the computing device, a model of the behavior pattern expressed in a vector format, based on a number of times that each of the extracted N-gram behavior sequences occurs;

comparing at least one of the generated model of the behavior pattern or the stored model of the behavior pattern with a reference model

determining, based on the comparison, whether the executed instruction set is malicious or normal.

5. The method of claim 4, wherein the hooking comprises dynamically hooking the system calls while the instruction set is executed in the computing device.

6. The method of claim 4, wherein the vector format is obtained by associating some or all of behavior sequences expressed in the N-gram manner with the number of times that each of the extracted behavior sequences occurs.