Granted Patent B2

US 12,609,969 · App. 17/980,000 · Granted Apr 21, 2026

Inventors: Radoslaw Gasiorek (Edinburgh, GB); John M. Nicholas (Fairfax, VA); Raul Garcia Calvo (Seville, ES); William Brad Arndt (Kensington, MD); Ryan Marcotte (Providence, RI)

Assignee: Secureworks Corp.

H04L63/205H04L63/1425H04L63/1441

Abstract

Claims (57)

1 . A system for detecting security threats, comprising:

a memory; and

at least one processor comprising:

a profiler engine configured to:

obtain data at rest from a repository and corresponding to a plurality of entities, mine selected statistical information from the obtained data at rest and generate a plurality of statistical profiles based on the obtained data at rest, the statistical profiles comprising the selected statistical information relating to each entity of the plurality of entities; and

store the plurality of statistical profiles in a profile database; and

a detector engine configured to:

generate one or more low fidelity detectors based on the statistical profiles of each entity of the plurality of entities organized in one or more threat chains, the one or more low fidelity detectors configured to detect a plurality of low level security threats in one or more threat chains, wherein each of the one or more threat chains comprises a sequence of events defining a security threat, wherein an individual low level security threat in a threat chain constitutes a non-actionable alert condition;

monitor, via each of the one or more low fidelity detectors, real-time data corresponding to the plurality of entities for the plurality of low level security threats, generate a score based on a correlation of a detected aggregation of the plurality of low level security threats to the plurality of statistical profiles, wherein the score is further based on an occurrence amount of low level security threats in one of the one or more threat chains; and

if the score exceeds a threshold, generate an alert.

2 . The system of claim 1 , wherein the obtained data at rest comprises historical data, and wherein the repository comprises a data lake.

3 . The system of claim 1 , wherein the plurality of entities comprise one or more of a user or a computing device.

4 . The system of claim 1 , wherein the one or more low fidelity detectors are respectively configured to detect a sequence, subsequence, low level security threat, or a combination thereof of one of the threat chains.

5 . The system of claim 1 , wherein the score is further based on an order in which low level security threats occur.

6 . The system of claim 1 , wherein each of the plurality of statistical profiles includes statistical analysis for a corresponding entity.

7 . The system of claim 6 , wherein one or more of the plurality of statistical profiles include statistics corresponding to a service accessed by the corresponding entity.

8 . The system of claim 1 , the at least one processor further comprising a normalization engine configured to:

receive data from a plurality of different data sources; and

normalize the received data prior to generation of the plurality of statistical profiles and storage thereof in the profile database.

9 . A computer-implemented method for detecting security threats, the method comprising:

receiving security data associated with one or more entities from a plurality of data sources;

normalizing the security data to generate normalized data;

providing the normalized data to a profiler engine;

generating, via the profiler engine, one or more statistical profiles for the one or more entities based on the normalized data;

generating one or more detectors based on one or more subsequences of low level security threats organized in a plurality of threat chains, wherein an individual low level security threat in a threat chain constitutes a non-actionable alert condition;

monitoring, via the one or more detectors, telemetric data in real time for the one or more subsequences for each of the plurality of threat chains;

in response to detection of one or more subsequences of one of the plurality of threat chains by the one or more detectors, aggregating the detected one or more subsequences;

generating a score based on a correlation of the aggregated detected one or more subsequences to the one or more statistical profiles, wherein the score is further based on an order of occurrence of the low level security threats in the one or more subsequences in a corresponding threat chain of the plurality of threat chains; and

if the score exceeds a threshold score, generating a high severity alert.

10 . The computer-implemented method of claim 9 , further comprising, prior to generating the one or more statistical profiles, obtaining the security data from a data lake, wherein the security data includes historical data.

11 . The computer-implemented method of claim 9 , wherein the one or more subsequences comprise one or more of events, files associated with the events, time stamps, or identifying data.

12 . The computer-implemented method of claim 11 , wherein the identifying data comprises one or more of usernames or IP addresses.

13 . The computer-implemented method of claim 9 , further comprising storing the one or more statistical profiles in a profile database.

14 . The computer-implemented method of claim 9 , wherein normalizing normalizes the security data based on a command associated with the security data and identified based on nomenclature associated with the command.

15 . The computer-implemented method of claim 9 , wherein normalizing normalizes the security data based on a unified format for the security data in a device agnostic manner.

16 . A non-transitory machine-readable storage medium with instructions stored thereon that, responsive to execution by a processing device, cause the processing device to perform or control performance of operations comprising:

in response to reception of security data associated with one or more entities from a plurality of data sources, normalizing the security data and storing the normalized security data;

generating one or more statistical profiles for the one or more entities based on the stored normalized security data;

generating one or more detectors based on one or more subsequences organized in a plurality of threat chains of low level security threats, wherein an individual low level security threat in a threat chain constitutes a non-actionable alert condition;

monitoring, via the one or more detectors, telemetric data from a cloud-based platform in real time for the one or more subsequences for each of the plurality of threat chains;

in response to detection of one or more subsequences of one of the plurality of threat chains by the one or more detectors, aggregating the detected one or more subsequences;

generating a score based on a correlation of the aggregated detected one or more subsequences to the one or more statistical profiles, wherein the score is further based on an order of occurrence of the low level security threats in the one or more subsequences in a corresponding threat chain of the plurality of threat chains; and

if the score exceeds a threshold score, generating a high severity alert.

17 . The non-transitory machine-readable storage medium of claim 16 , wherein the one or more entities comprise one or more users.

18 . The non-transitory machine-readable storage medium of claim 17 , wherein generating the one or more statistical profiles for the one or more entities comprises conducting a statistical analysis of the stored normalized security data; and

wherein a result of the statistical analysis of the stored normalized security data comprises a number of times each of the one or more users has accessed a cloud-based service on the cloud-based platform.

19 . A computer-implemented method for detecting security threats, the method comprising:

storing historical telemetric data at rest in a data lake;

building statistical profiles based on the historical telemetric data at rest in the data lake;

receiving one or more definitions for one or more low fidelity detectors configured to detect different low level security threats in one or more threat chains, wherein each of the one or more threat chains comprises a sequence of events defining a security threat, wherein an individual low level security threat in a threat chain constitutes a non-actionable alert condition;

generating the one or more low fidelity detectors based on the one or more definitions;

listening to real-time telemetric data via the one or more low fidelity detectors;

in response to detection of low level security threats by the one or more low fidelity detectors, correlating the low level security threats with the statistical profiles to generate a score, wherein the score is further based on an occurrence amount of low level security threats in one of the one or more threat chains; and

in response to the score exceeding a threshold, generating a high severity alert.

20 . The computer-implemented method of claim 19 , further comprising:

reviewing and refining the one or more low fidelity detectors with a security computing device; and

transmitting the high severity alert to the security computing device or another security computing device.

Assignments (2)

Continuity (1)

References Cited (400)