Modeling and outlier detection in threat management system data
Methods, systems, and computer-readable media for identifying potential threats on a network based on anomalous behavior in communication between endpoints are provided. Traffic data for a network is accumulated over some period of time. The traffic data is grouped by one or more keys, such as source IP address, and sets of metric values are calculated for the keys. A mixture distribution, such as a negative binomial mixture distribution, is fitted to each set of metric values, and outlying metric values are determined based on the mixture distribution(s). A list of outliers is then generated comprising key values having outlying metric values in one or more of the sets of metric values.
1. A method of identifying potential threats on a network comprising:
accumulating, by a processor, traffic data from the network over a period of time;
calculating, by the processor, a first set of metric values for endpoints communicating on the network from the traffic data;
fitting, by the processor, a first mixture distribution to the first set of metric values;
identifying, by the processor, outlying metric values based on the mixture distribution; and
generating, by the processor, a list of outliers comprising the endpoints having an outlying metric value, wherein a second set of metric values are calculated for the endpoints from the traffic data, wherein a second mixture distribution is fitted to the second set of metric values, and wherein the list of outliers comprises the endpoints having outlying metric values in the first set of metric values and the second set of metric values.
2. The method of claim 1 , further comprising clustering the endpoints in the list of outliers based on similar communication behavior patterns exhibited by the endpoints.
3. The method of claim 2 , further comprising co-clustering across the endpoints with different key values based on a relationship between the endpoints represented in the traffic data.
4. The method of claim 1 , wherein the first mixture distribution comprises a negative binomial mixture distribution.
5. The method of claim 1 , wherein identifying the outlying metric values comprises calculating a fit quantity for each metric value in the first set of metric values indicating if the metric value would be generated by a high probability from dominant components of the first mixture distribution, and identifying the metric values having the fit quantity below a specific threshold as outliers.
6. The method of claim 1 , further comprising calculating a perplexity of the first set of metric values and identifying outliers based on the perplexity of each metric value in the first set of metric values.
7. A system comprising:
a memory for storing a program containing computer-executable instructions for identifying potential threats on a network; and
a processor being responsive to the computer-executable instructions the, when executed by the processor, cause the processor to perform operations comprising
receiving traffic data accumulated from the network over a period of time,
calculating a first set of metric values for endpoints communicating on the network from the traffic data,
calculating a second set of metric values for the endpoints from the traffic data,
fitting a first mixture distribution to the first set of metric values,
fitting a second mixture distribution to the second set of metric values,
identifying outlying metric values based on the first mixture distribution,
generating a list of outliers comprising the endpoints having outlying metric values, wherein generating the list of outliers comprising the endpoints having outlying metric values comprises generating the list of outliers comprising the endpoints having outlying metric values in the first set of metric values and the second set of metric values.
8. The system of claim 7 , wherein the operations further comprise clustering the endpoints in the list of outliers based on similar communication behavior patterns exhibited by the endpoints.
9. The system of claim 8 , wherein the operations further comprise co-clustering across the endpoints with different key values based on a relationship between the endpoints represented in the traffic data.
10. The system of claim 7 , wherein the first mixture distribution comprises a negative binomial mixture distribution.
11. The system of claim 7 , wherein identifying the outlying metric values comprises calculating a fit quantity for each metric value in the first set of metric values indicating if the metric value would be generated by a high probability from dominant components of the first mixture distribution, and identifying the metric values having the fit quantity below a specific threshold as outliers.
12. The system of claim 7 , wherein the operations further comprise calculating a perplexity of the first set of metric values and identifying outliers based on the perplexity of each metric value in the first set of metric value.
13. A computer-readable storage medium having computer-executable instructions stored thereon that, when executed by a computer, cause the computer to perform operations comprising:
receiving traffic data accumulated from a network;
calculating a first set of metric values for endpoints communicating on the network from the traffic data;
calculating a second set of metric values for the endpoints from the traffic data;
fitting a negative binomial mixture distribution to the first set of metric values;
fitting a mixture distribution to the second set of metric values;
identifying outlying metric values based on the negative binomial mixture distribution; and
generating a list of outliers comprising the endpoints having an outlying metric value, wherein generating the list of outliers comprising the endpoints having the outlying metric value comprises generating the list of outliers comprising endpoints having outlying metric values in the first sets of metric values and the second set of metric values.
14. The computer-readable storage medium of claim 13 , wherein the operations further comprise clustering the endpoints in the list of outliers based on similar communication behavior patterns exhibited by the endpoints.
15. The computer-readable storage medium of claim 14 , wherein the operations further comprise co-clustering across the endpoints with different key values based on a relationship between the endpoints represented in the traffic data.
16. The computer-readable storage medium of claim 13 , wherein identifying the outlying metric values comprises calculating a fit quantity for each metric value in the first set of metric values indicating if the metric value would be generated by dominant components of the negative binomial mixture distribution and identifying the metric values having the fit quantity below a specific threshold as outliers.
17. The computer-readable storage medium of claim 13 , wherein the operations further comprise calculating a perplexity of the first set of metric values and identifying outliers based on the perplexity of each of the first set of metric values.