DYNAMIC ACTIVITY MODEL OF NETWORK SERVICES

• US 20080101352A1
• Filed: 10/31/2006
• Published: 05/01/2008
• Est. Priority Date: 10/31/2006
• Status: Active Grant

• Current Assignee: Zhigu Holdings Limited
• Original Assignee: Microsoft Corporation
• Inventors: Goldszmidt, Moises, Black, Richard, Mortier, Richard, MacCormick, John, Isaacs, Rebecca, Barham, Paul
• Granted Patent: US 7,949,745 B2
• Time in Patent Office: Days
• US Class Current: 370/389
• CPC Class Codes: H04L 41/0645 by additionally acting on or stimulating the network after receiving notifications
• CPC Class Codes: H04L 41/142 using statistical or mathematical methods
• CPC Class Codes: H04L 41/16 using machine learning or artificial intelligence

Abstract

An activity model is generated at a computer. The activity model may be generated by monitoring incoming and outgoing data in the computer. The collected data is analyzed to form a graph that describes and predicts what output is generated in response to received input. Later, a window of input and output data is collected from the computer. This collected window of data is used to query the activity model. The graph in the activity model is then used to give the probability that the collected window of data was collected from the computer used to generate the activity model. A high probability indicates that the computer is performing normally, while a low probability indicates that the computer may behaving erratically and there may be a problem with the computer.

Claims

1. A method for constructing and maintaining an activity model, comprising:

capturing a first window of input and output data into a computer;

analyzing the collected first window of input and output data to identify statistical relationships among the data; and

generating an activity model from the identified statistical relationships among the first window of data.

2. The method of claim 1, further comprising:

capturing a second window of input and output data into the computer; and

querying the activity model using the captured second window of data to determine the likelihood that the computer is behaving normally.

3. The method of claim 1, further comprising continuously capturing subsequent windows of input and output data and updating the generated activity model to account for the captured windows of input and output data.

4. The method of claim 1, wherein the activity model comprises a graph.

5. The method of claim 4, wherein the graph comprises a noisy-or graph.

6. The method of claim 4, wherein the graph comprises a Bipartite Bayesian Network.

7. The method of claim 4, wherein the graph comprises an Arbitrary Graph Structure.

8. The method of claim 4, wherein the graph comprises a Bayesian classifier model.

9. The method of claim 1, wherein the data is packets.

10. The method of claim 9, wherein analyzing the collected first window of input and output data to identify statistical relationships among the data comprises analyzing origin and destination addresses of the packets.

11. The method of claim 9, wherein analyzing the collected first window of input and output data to identify statistical relationships among the data comprises analyzing service information associated with the packets.

12. A computer-readable medium containing computer-executable instructions for performing the method of:

selecting a first window of input and output data;

analyzing the collected first window of input and output data to identify statistical relationships among the data; and

generating an activity model from the identified statistical relationships among the first window of data.

13. The computer-readable medium of claim 12, further comprising computer-executable instructions for:

selecting a second window of input and output data; and

querying the activity model using the captured second window of data to determine if the captured second window of data is predicted by the generated activity model.

14. The computer-readable medium of claim 12, further comprising computer-executable instructions for continuously selecting subsequent windows of input and output data and updating the generated activity model to account for the subsequent windows of input and output data.

15. The computer-readable medium of claim 12, wherein the activity model comprises a graph.

16. The computer-readable medium of claim 15, wherein the graph comprises one of a noisy- or graph, a Bipartite Bayesian Network, a Bayesian classifier model, and an Arbitrary Graph Structure.

17. The computer-readable medium of claim 12, wherein the data is packets.

18. The computer-readable medium of claim 17, wherein analyzing the selected first window of input and output data to identify statistical relationships among the data comprises computer-executable instructions for analyzing origin and destination addresses of the packets.

19. The computer-readable medium of claim 17, wherein analyzing the selected first window of input and output data to identify statistical relationships among the data comprises computer-executable instructions for analyzing service information associated with the packets.

20. A method for constructing an activity model, comprising:

capturing a first window of input and output packets into a computer;

analyzing the arrival time and the source and destination addresses of the packets to identify statistical relationships among the packets; and

generating an activity model from the identified statistical relationships among the packets.