Honeypot network management based on probabilistic detection of malicious port activity
Disclosed herein are methods, systems, and processes for probabilistically identifying anomalous levels of honeypot activity. A honeypot dataset associated with a honeypot network is received and a representative usage value is determined from the honeypot dataset. The representative usage value is identified as being associated with anomalous behavior if the representative usage value deviates from an expected probability distribution. A remediation operation is initiated in the honeypot network in response to the identification of the representative usage value as being associated with the anomalous behavior by virtue of the representative usage value deviating from the expected probability distribution.
1. A computer-implemented method comprising:
receiving a honeypot dataset associated with a honeypot network, wherein the honeypot dataset comprises one or more connections to at least one or more ports of one or more honeypots in the honeypot network over a period of time;
determining a number of connections to a port of the one or more ports over the period of time;
determining that the number of connections is associated with malicious activity based on a probability of a deviation of the number of connections from a historical average of the number of connections based on an expected probability distribution according to a discrete probability distribution function configured for the port;
identifying one or more malicious computing devices associated with the malicious activity;
modifying one or more intrusion detection rules or intrusion alert triggers based on the malicious activity and the one or more malicious computing devices; and
performing a remediation operation on one or more computer devices in response to the determination that the number of connections is associated with malicious activity, wherein the remediation operation includes one or more of preventing access to or exfiltration from the one or more computer devices or generating a security workflow for vulnerability assessment of the one or more computer devices.
2. The computer-implemented method of claim 1 , wherein the expected probability distribution comprises a Negative Binomial distribution, the expected probability distribution comprises a Poisson distribution, and the determination that the number of connections is associated with malicious activity is made based on an activity level probability of a representative usage value of the port being lower than a projected activity probability estimated by the Poisson distribution.
3. The computer-implemented method of claim 1 , wherein the historical average is determined from one or more previous honeypot datasets received from the honeypot network.
4. The computer-implemented method of claim 1 , wherein the identification of the one or more malicious computing devices and the modification of the one or more intrusion detection rules or intrusion alert triggers are performed by an anomaly detection and remediation (ADR) server.
5. The computer-implemented method of claim 4 , wherein the honeypot network includes virtual machines hosted in a cloud network.
6. The computer-implemented method of claim 5 , wherein the ADR server is a virtual machine hosted in the cloud network.
7. The computer-implemented method of claim 1 , wherein the honeypot dataset is received over a public network.
8. The computer-implemented method of claim 1 , further comprising:
monitoring for connection anomalies of the port based on observed connection durations of the port.
9. The computer-implemented method of claim 1 , further comprising:
monitoring for connection anomalies to a plurality of ports in the honeypot network, wherein each port is monitored based on a different discrete probability function.
10. The computer-implemented method of claim 9 , further comprising:
generating anomaly events for individual ports based on the monitoring;
ranking the anomaly events; and
using the ranking to prioritize remediation operations on the ports.
11. The computer-implemented method of claim 10 , wherein the ranking comprises (a) a level ranking based on a number of anomaly events observed for a particular port, and (b) a scale ranking based on a distribution of anomaly events observed over multiple honeypots in the honeypot network.
12. The computer-implemented method of claim 10 , wherein the remediation operation is prioritized based on the ranking of anomaly events.
13. A system comprising:
one or more computing devices that implement an anomaly detection and remediation (ADR) server, configured to:
receive a honeypot dataset associated with a honeypot network, wherein the honeypot dataset comprises one or more connections to at least one or more ports of one or more honeypots in the honeypot network over a period of time;
determine a number of connections to a port of the one or more ports over the period of time;
determine that the number of connections is associated with malicious activity based on a probability of a deviation of the number of connections from a historical average of the number of connections based on an expected probability distribution according to a discrete probability distribution function configured for the port;
identify one or more malicious computing devices associated with the malicious activity;
modify one or more intrusion detection rules or intrusion alert triggers based on the malicious activity and the one or more malicious computing devices; and
perform a remediation operation on one or more computer devices in response to the determination that the number of connections is associated with malicious activity, wherein the remediation operation includes one or more of preventing access to or exfiltration from the one or more computer devices or generating a security workflow for vulnerability assessment of the one or more computer devices.
14. The system of claim 13 , wherein the expected probability distribution comprises a Negative Binomial distribution, the expected probability distribution comprises a Poisson distribution, and the determination that the number of connections is associated with malicious activity is made based on an activity level probability of a representative usage value of the port being lower than a projected activity probability estimated by the Poisson distribution.
15. The system of claim 13 , wherein the historical average is determined from one or more previous honeypot datasets received from the honeypot network.
16. The system of claim 13 , wherein the honeypot network includes virtual machines hosted in a cloud network, and the honeypot dataset is received over a public network.
17. The system of claim 13 , wherein the ADR server is configured to:
monitor for connection anomalies of the port based on observed connection durations of the port.
18. The system of claim 13 , wherein the ADR server is configured to:
monitor for connection anomalies to a plurality of ports in the honeypot network, wherein each port is monitored based on a different discrete probability function.
19. The system of claim 18 , wherein the ADR server is configured to:
generate anomaly events for individual ports based on the monitoring;
rank the anomaly events; and
use the ranking to prioritize remediation operations on the ports.
20. The system of claim 19 , wherein the remediation operation is prioritized based on the ranking of anomaly events.