Rogue base station router detection with machine learning algorithms
This application is directed to a method for detecting a rogue device in a network. The method includes a step of surveying the network. The method also includes a step of collecting broadcast data from cellular towers in the network based on the survey. The method also includes a step of distilling the collected broadcast data into abstract syntax notation one (ASN.1)-encoded system information blocks (SIBs) associated with plural devices. The method further includes a step of featurizing the ASN.1-encoded SIBs. The method even further includes a step of running the featurized, ASN.1-encoded SIBs through an unsupervised machine learning algorithm. The algorithm is executed by a processor to analyze all cells in the survey for the rogue device. Yet even further, the method includes a step of determining, based on the run, anomalous cells exhibiting characteristics of the rogue device from all cells in the survey.
1. A method for detecting a rogue device in a network comprising:
surveying the network;
collecting, based on the survey, broadcast data from cellular towers in the network;
distilling the collected broadcast data into abstract syntax notation one (ASN.1)-encoded system information blocks (SIBs) associated with plural devices;
featurizing the ASN.1-encoded SIBs;
determining, for each of the plural devices, the times of a first and a last occurrence of the device on the network;
running the featurized, ASN.1-encoded SIBs and the respective times of the first and last occurrences of the plural devices through an unsupervised machine learning algorithm, executed by a processor, to analyze all cells in the survey for the rogue device; and
determining, based on the run, anomalous cells exhibiting characteristics of the rogue device from all cells in the survey.
2. The method of claim 1 , wherein the collected broadcast data is based on a single radio access technology (RAT).
3. The method of claim 2 , wherein the collected broadcast data is based on a public land mobile network (PLMN).
4. The method of claim 2 , wherein the RAT is UMTS or LTE.
5. The method of claim 1 , wherein the machine learning algorithm is selected from a k-means clustering algorithm, an isolation forest anomaly detection algorithm, and an agglomerative hierarchical clustering algorithm.
6. The method of claim 5 , wherein the ASN.1-encoded SIBs include one or more of Universal Mobile Telecommunications System (UMTS) SIB 1, UMTS SIB 3 and UMTS SIB 5.
7. The method of claim 5 , wherein the ASN.1-encoded SIBs include one or more of Long-Term Evolution (LTE) SIB 1, LTE SIB 2 and LTE SIB 3.
8. The method of claim 1 , wherein
the determining step includes confirming a number of native cells of a commercial carrier is greater than a number of rogue cells, and
the determined cell is based on a single cellular scan of a geographic area.
9. The method of claim 1 , further comprising notifying users on the network of the rogue device.
10. The method of claim 1 , wherein the determined rogue device is a cellular tower or dynamic base station router.
11. The method of claim 1 , further comprising:
displaying, on a graphical user interface (GUI), a location of the rogue device in the network.
12. The method of claim 1 , further comprising:
displaying, on a GUI, the respective times of the first and last occurrences of the plural devices on the network; and
notifying users on the network of the rogue device based on the respective times of the first and last occurrences of the plural devices on the network.
13. The method of claim 1 , further comprising:
identifying a manufacturer, technologies in use, or combinations thereof of the rogue device.
14. The method of claim 13 , wherein the identifying step is based on use of at least one of a neural network and a support vector machine.