Creating aggregate network flow time series in network anomaly detection systems
In an embodiment, a computer implemented method receives flow data for one or more flows that correspond to a device-circuit pair. The method calculates a time difference for each flow that corresponds to a device-circuit pair. Based on the calculated time differences and the received flow data, the method updates a probability distribution model associated with the device-circuit pair. Then, the method determines whether a time bucket is complete or open based on the updated probability distribution model.
1. A computer implemented method for processing a network flow with a device-circuit pair, comprising:
obtaining, by a server, a probability distribution model associated with the device-circuit pair, the probability distribution model indicating, for different time differences of prior network flows through the device-circuit pair, corresponding numbers of occurrences;
determining, by the server, a duration of a time bucket according to the probability distribution model associated with the device-circuit pair including determining, by the server, a time delay value according to the probability distribution model and extending, by the server, an end time of the time bucket according to the time delay value;
determining, by the server, whether to ignore the network flow according to the duration of the time bucket; and
detecting, by the server, a network anomaly associated with the device-circuit pair according to the determination of whether to ignore the network flow, wherein
each of the time differences is a difference between i) a start time or an end time of a corresponding prior network flow at one of the device-circuit pair and ii) a file stamp time of the corresponding prior network flow received by the server.
2. The method of claim 1 , wherein determining the time delay value according to the probability distribution model includes:
determining, by the server, the time delay value as a function of a standard deviation value and a mean value of the probability distribution model.
3. The method of claim 1 , wherein determining whether to ignore the network flow according to the duration of the time bucket includes:
determining, by the server, to ignore the network flow, in response to the network flow received by the server after the time bucket.
4. The method of claim 3 , wherein detecting the network anomaly associated with the device-circuit pair according to the determination of whether to ignore the network flow includes:
detecting the network anomaly associated with the device-circuit pair without the network flow.
5. The method of claim 1 , wherein determining whether to ignore the network flow according to the duration of the time bucket includes:
determining, by the server, to incorporate the network flow, in response to the network flow received by the server within the time bucket.
6. The method of claim 5 , wherein detecting the network anomaly associated with the device-circuit pair according to the determination of whether to ignore the network flow includes:
detecting the network anomaly associated with the device-circuit pair with the network flow.
7. A system for processing a network flow with a device-circuit pair, comprising:
one or more processors; and
a non-transitory computer readable storing instructions when executed by the one or more processors cause the one or more processors to:
obtain a probability distribution model associated with the device-circuit pair, the probability distribution model indicating, for different time differences of prior network flows through the device-circuit pair, corresponding numbers of occurrences,
determine a duration of a time bucket according to the probability distribution model associated with the device-circuit pair including determining a time delay value according to the probability distribution model and extending an end time of the time bucket according to the time delay value,
determine whether to ignore the network flow according to the duration of the time bucket, and
detect a network anomaly associated with the device-circuit pair according to the determination of whether to ignore the network flow, wherein
each of the time differences is a difference between i) a start time or an end time of a corresponding prior network flow at one of the device-circuit pair and ii) a file stamp time of the corresponding prior network flow received by the server.
8. The system of claim 7 , wherein the instructions that cause the one or more processors to determine the time delay value according to the probability distribution model include instructions when executed by the one or more processors cause the one or more processors to:
determine the time delay value as a function of a standard deviation value and a mean value of the probability distribution model.
9. The system of claim 7 , wherein the instructions that cause the one or more processors to determine whether to ignore the network flow according to the duration of the time bucket include instructions when executed by the one or more processors cause the one or more processors to:
determine to ignore the network flow, in response to the network flow received by the system after the time bucket.
10. The system of claim 9 , wherein the instructions that cause the one or more processors to detect the network anomaly associated with the device-circuit pair according to the determination of whether to ignore the network flow include instructions when executed by the one or more processors cause the one or more processors to:
detect the network anomaly associated with the device-circuit pair without the network flow.
11. The system of claim 7 , wherein the instructions that cause the one or more processors to determine whether to ignore the network flow according to the duration of the time bucket include instructions when executed by the one or more processors cause the one or more processors to:
determine to incorporate the network flow, in response to the network flow received by the system within the time bucket.
12. The system of claim 11 , wherein the instructions that cause the one or more processors to detect the network anomaly associated with the device-circuit pair according to the determination of whether to ignore the network flow include instructions when executed by the one or more processors cause the one or more processors to:
detect the network anomaly associated with the device-circuit pair with the network flow.
13. A non-transitory computer readable medium for processing a network flow with a device-circuit pair, the non-transitory computer readable medium storing instructions when executed by one or more processors cause the one or more processors to:
obtain a probability distribution model associated with the device-circuit pair, the probability distribution model indicating, for different time differences of prior network flows through the device-circuit pair, corresponding numbers of occurrences;
determine a duration of a time bucket according to the probability distribution model associated with the device-circuit pair including determining a time delay value according to the probability distribution model and extending an end time of the time bucket according to the time delay value;
determine whether to ignore the network flow according to the duration of the time bucket; and
detect a network anomaly associated with the device-circuit pair according to the determination of whether to ignore the network flow, wherein
each of the time differences is a difference between i) a start time or an end time of a corresponding prior network flow at one of the device-circuit pair and ii) a file stamp time of the corresponding prior network flow received by the server.
14. The non-transitory computer readable medium of claim 13 , wherein the instructions that cause the one or more processors to determine the time delay value according to the probability distribution model include instructions when executed by the one or more processors cause the one or more processors to:
determine the time delay value as a function of a standard deviation value and a mean value of the probability distribution model.
15. The non-transitory computer readable medium of claim 13 , wherein the instructions that cause the one or more processors to determine whether to ignore the network flow according to the duration of the time bucket include instructions when executed by the one or more processors cause the one or more processors to:
determine to ignore the network flow, in response to the network flow received by a server after the time bucket.