Non-harmful insertion of data mimicking computer network attacks

Patent metadata

Field	Value
Publication number	US-10122741-B2
Application number	US-201615165036-A
Country	US
Kind code	B2
Filing date	May 26, 2016
Priority date	Mar 22, 2012
Publication date	Nov 6, 2018
Grant date	Nov 6, 2018

Abstract

Non-harmful data mimicking computer network attacks may be inserted in a computer network. Anomalous real network connections may be generated between a plurality of computing systems in the network. Data mimicking an attack may also be generated. The generated data may be transmitted between the plurality of computing systems using the real network connections and measured to determine whether an attack is detected.

First claim

The invention claimed is:

1. A computer-implemented method, comprising:
determining based on historical data, by a computing system, which anomalous real network connections to generate between a plurality of computing systems in a network to create the anomalous real network connections based on stochastic models, the anomalous real network connections forming a k-path between the plurality of computing systems comprising a plurality of directed edges, each directed edge comprising associated data;
generating, by the plurality of computing systems, the determined anomalous real network connections between the plurality of computing systems in the network and data mimicking an attack; and
transmitting, by the plurality of computing systems, the generated data mimicking the attack between the plurality of computing systems using the determined anomalous real network connections forming respective k-paths, wherein
the k-path is a subgraph of size k that has diameter k, and
k is at least two.

2. The computer-implemented method of claim 1 , wherein the generated data comprises Domain Name Server (“DNS”) traffic that is transmitted in a path comprising computing systems that have never communicated before.

3. The computer-implemented method of claim 1 , wherein the data mimicking the attack is designed to mimic one or more specific types of attacks based on known types of real attacks.

4. The computer-implemented method of claim 1 , further comprising:
building stochastic models, by the computing system, for each directed edge in k-paths of a plurality of the anomalous network connections; and
comparing historical parameters with current parameters in a time window, by the computing system, to determine a level of anomalousness of the k-paths of the plurality of the anomalous network connections.

5. The computer-implemented method of claim 4 , wherein the historical data comprises historical connection patterns between the plurality of computing systems.

6. The computer-implemented method of claim 1 , further comprising:
collecting, by the computing system, the generated data mimicking the attack;
analyzing, by the computing system, the collected data;
determining, by the computing system, whether an attack has occurred based on the collected data; and
verifying, by the computing system, that the attack was correctly detected.

7. The computer-implemented method of claim 6 , further comprising:
providing output of results, by the computing system, indicating effectiveness of the verification.

8. A computer program embodied on a non-transitory computer-readable medium, the computer program configured to cause at least one processor to:
determine, based on historical data, which network connections to generate that form a k-path between a plurality of computing systems to create an anomaly by applying stochastic models for each edge in the k-path;
generate the determined network connections in the k-path and data mimicking an attack to create the anomaly; and
transmit the generated data mimicking the attack between the plurality of computing systems using the determined anomalous real network connections forming the k-path, wherein
the k-path is a subgraph of size k that has diameter k, and
k is at least two.

9. The computer program of claim 8 , the program further configured to cause the at least one processor to:
verify that a data collection mechanism for the network actually measured data pertaining to the generated connections; and
provide output of results indicating a degree of success or failure of the verification.

10. The computer program of claim 8 , wherein the network anomaly is designed to mimic one or more specific types of attacks based on known types of real attacks.

11. The computer program of claim 8 , wherein the generated data comprises Domain Name Server (“DNS”) traffic that is transmitted in a path comprising computing systems that have never communicated before.

12. The computer program of claim 8 , the program further configured to cause the at least one processor to:
build stochastic models for each directed edge in the k-paths of a plurality of the determined network connections; and
compare historical parameters with current parameters in a time window to determine a level of anomalousness of the k-paths of the plurality of the determined network connections.

13. The computer program of claim 8 , wherein the historical data comprises historical connection patterns between the plurality of computing systems.

14. A computer-implemented method, comprising:
inserting, by a plurality of computing systems, traffic in a k-path in a network comprising computing systems that do not normally communicate, the traffic deviating from stochastic models for each edge of the k-path;
detecting the inserted k-path as anomalous, by a computing system, based on analysis of the inserted traffic;
verifying, by the computing system, that an attack was correctly detected based on the detected inserted k-path; and
providing output of results, by the computing system, indicating a degree of success or failure of the verification, wherein
the k-path comprises a plurality of directed edges, each directed edge comprising associated data,
the k-path is a subgraph of size k that has diameter k, and
k is at least two.

15. The computer-implemented method of claim 14 , wherein the inserted traffic is designed to mimic one or more specific types of attacks based on known types of real attacks.

16. The computer-implemented method of claim 14 , further comprising:
building stochastic models, by the computing system, for each directed edge in the k-path; and
comparing historical parameters with current parameters in a time window, by the computing system, to determine a level of anomalousness of the k-path.

17. The computer-implemented method of claim 14 , further comprising:
determining based on historical data, by the computing system, which traffic to insert between the computing systems to create an anomaly, wherein
the historical data comprises historical connection patterns between the plurality of computing systems.

18. A computer program embodied on a non-transitory computer-readable medium, the computer program configured to cause at least one processor to:
determine, based on historical data, which network connections to generate that form a k-path between a plurality of computing systems to create an anomaly by applying stochastic models for each edge in the k-path;
generate the determined network connections in the k-path to create the anomaly;
verify that a data collection mechanism for the network actually measured data pertaining to the generated connections; and
provide output of results indicating a degree of success or failure of the verification, wherein
the k-path is a subgraph of size k that has diameter k, and
k is at least two.

19. The computer program of claim 18 , wherein the network anomaly is designed to mimic one or more specific types of attacks based on known types of real attacks.

20. The computer program of claim 18 , wherein the generated data comprises Domain Name Server (“DNS”) traffic that is transmitted in a path comprising computing systems that have never communicated before.

21. The computer program of claim 18 , the program further configured to cause the at least one processor to:
build stochastic models for each directed edge in the k-paths of a plurality of the determined network connections; and
compare histori

Assignees

• Los Alamos Nat Security Llc

Inventors

• Neil Joshua
• Kent Alexander
• Hash Jr Curtis
• Fisk Michael
• Brugh Alexander
• Storlie Curtis
• Uphoff Benjamin

Classifications

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G06N5/022

  Knowledge engineering; Knowledge acquisition · CPC title

• G06N5/045

  Explanation of inference; Explainable artificial intelligence [XAI]; Interpretable artificial intelligence · CPC title

• H04L2463/144

  Detection or countermeasures against botnets · CPC title

• H04L63/1433

  Vulnerability analysis · CPC title