×

Probabilistic cyber threat recognition and prediction

  • US 9,367,694 B2
  • Filed: 05/16/2014
  • Issued: 06/14/2016
  • Est. Priority Date: 05/16/2014
  • Status: Active Grant
First Claim
Patent Images

1. A method for recognizing a cyber threat comprising:

  • determining, using a processor, a network layout of a network based on received network layout data, the network layout data indicating connections between a plurality of nodes of the network;

    receiving cyber sensor data indicating actions performed on the plurality of nodes of the network;

    determining that the cyber sensor data does not sufficiently match known cyber threat profiles;

    determining, using an Interacting Multiple Model (IMM), one or more hybrid cyber threat profiles, the hybrid cyber threat profiles including a combination of portions of two or more of the known cyber threat profiles, each of the hybrid cyber threat profiles and the known cyber threat profiles indicating evidence left behind by an associated cyber threat and hybrid cyber threat, respectively, wherein the evidence includes two or more of a security or application log being created, a login attempt, a file or program being accessed, a program being run, determining a layout of at least a portion of the network, scanning, enumeration, gaining access to the network, escalating a user'"'"'s privilege status;

    calculating, using the processor, a first score associated with the cyber sensor data indicating that the hybrid cyber threat is present in the network by comparing the cyber threat profile of the hybrid cyber threat to actions performed on the network as indicated by the cyber sensor data; and

    determining that the hybrid cyber threat is present in response to determining the calculated first score is greater than a specified threshold.

View all claims
  • 1 Assignment
Timeline View
Assignment View
    ×
    ×