Data exfiltration detector
Methods and systems for detecting a data exfiltration event on a network. The method includes receiving traffic data and applying a transformation to transform the traffic data at least closer to a normal distribution. The method further includes selecting at least one outlier identification technique based on a property of the transformed data, and then executing the at least one selected identification technique to determine whether the traffic data is indicative of a data exfiltration event.
1. A method for detecting a data exfiltration event on a network, the method comprising:
receiving traffic data regarding outgoing traffic from a source associated with the traffic data;
applying a normalizing transformation to the received traffic data resulting in transformed data;
selecting at least one outlier identification technique based on a property of the transformed data;
executing the at least one selected outlier identification technique on the transformed data to determine whether the traffic data regarding the outgoing traffic is indicative of an exfiltration event; and
performing at least one remedial action upon determining the traffic data regarding the outgoing traffic is indicative of an exfiltration event.
2. The method of claim 1 wherein receiving the traffic data includes selecting a subset of network activity data, and the subset is selected based on at least one of time, timeframe, and source associated with the network activity data.
3. The method of claim 1 wherein applying a normalizing transformation includes performing a logarithmic transformation on the received traffic data.
4. The method of claim 1 wherein the property of the transformed data includes at least one of standard deviation and interquartile range.
5. The method of claim 1 wherein the determination of whether the traffic data regarding the outgoing traffic is indicative of an exfiltration event is based on a comparison against a baseline including at least one of time, duration, and source address, wherein the comparison is generated by the selected outlier identification technique.
6. The method of claim 1 further comprising applying at least one of:
a minimum size contextual filter to the results of the outlier identification technique so that only outgoing traffic data that exceeds a minimum size threshold is determined to be indicative of an exfiltration event, and
a magnitude contextual filter so that only outgoing traffic data that exceeds a threshold with respect to incoming traffic data is determined to be indicative of an exfiltration event.
7. The method of claim 1 further comprising applying a classifier to the results of the selected outlier identification technique to identify items of data indicative of a data exfiltration event on the network.
8. The method of claim 7 wherein the classifier is manually trained using feedback with respect to outgoing traffic indicative of an exfiltration event, and the method further includes implementing the feedback in future executions of the at least one outlier identification technique.
9. The method of claim 1 wherein the transformed data includes a plurality of uniform values, and the method further includes multiplying the uniform value by a constant to obtain a scaled value,
wherein the determination that the outgoing traffic is indicative of a data exfiltration event is based on the received traffic data exceeding the scaled value.
10. A system for detecting a data exfiltration event, the system comprising:
an interface for receiving traffic data regarding outgoing traffic from a source associated with the traffic data;
a processor executing instructions stored on a memory to:
apply a normalizing transformation to the received traffic data resulting in transformed data;
select at least one outlier identification technique based on a property of the transformed data;
execute the at least one selected outlier identification technique on the transformed data to determine whether the traffic data regarding the outgoing traffic is indicative of an exfiltration event; and
perform at least one remedial action upon determining the traffic data regarding the outgoing traffic is indicative of an exfiltration event.
11. The system of claim 10 wherein the processor is further configured to select a subset of network activity data as the traffic data regarding the outgoing traffic, and the processor selects the subset based on at least one of time, timeframe, and source associated with the network activity data.
12. The system of claim 10 wherein the processor applies the transformation by performing a logarithmic transformation on the received traffic data.
13. The system of claim 10 wherein the property of the transformed data includes at least one of standard deviation and interquartile range.
14. The system of claim 10 wherein the determination of whether the traffic data regarding the outgoing traffic is indicative of a data exfiltration event is based on a comparison against a baseline including at least one of time, duration, and source address, wherein the comparison is generated by the selected outlier identification technique.
15. The system of claim 10 wherein the processor is further configured to apply at least one of:
a minimum size contextual filter to the results of the outlier identification technique so that only outgoing traffic data that exceeds a minimum size threshold is determined to be indicative of an exfiltration event, and
a magnitude contextual filter so that only outgoing traffic data that exceeds a threshold with respect to incoming traffic data is determined to be indicative of an exfiltration event.
16. The system of claim 10 wherein the processor is further configured to apply a classifier to the results of the selected outlier identification technique to identify items of data indicative of a data exfiltration event on the network.
17. The system of claim 16 wherein the classifier is manually trained using feedback with respect to outgoing traffic indicative of an exfiltration event, and the processor is further configured to implement the feedback in future executions of the at least one outlier identification technique.
18. The system of claim 10 wherein the transformed data includes a plurality of uniform values, and the processor is further configured to multiply the uniform value by a constant to obtain a scaled value,
wherein the determination that the outgoing traffic is indicative of a data exfiltration event is based on the received traffic data exceeding the scaled value.
19. A method for detecting a data exfiltration event on a network, the method comprising:
receiving traffic data regarding outgoing traffic from a source associated with the traffic data;
performing a logarithmic transformation on the received traffic data resulting in transformed data;
calculating the interquartile range and the standard deviation of the transformed data;
selecting at least one outlier identification technique based on the calculated interquartile range and the standard deviation of the transformed data;
executing the at least one selected outlier identification technique on the transformed data to determine whether the traffic data regarding the outgoing traffic is indicative of an exfiltration event; and
performing at least one remedial action upon determining the traffic data regarding the outgoing traffic is indicative of an exfiltration event.