Abnormal Activity


Abnormal Account Activity

Monitoring use of user accounts is an effective way to identify suspicious activity.

  • e.g.,
    • A user account for an employee with well-defined working hours being used during the night
    • A user account provisioned to only work on desktop computers being used on a server computer
    • A user account created on a local computer or created by a user without authorization to create accounts
    • An account being added to a group unexpectedly or added by an unauthorized individual

Impossible travel is a tracking of information such as GPS address, IP address, or user’s device to pinpoint a user’s location and determine whether a behavior was physically possible.

  • an indicator that has become more common with remote work and cloud computing
  • could be result of VPN use

Abnormal Behavior and Patterns

Monitoring computer activity may reveal problematic activities or sequences of activities that match known attack patterns.

  • e.g.,
    • Evidence of communication with known malicious IP addresses or domain names
    • Abnormal communication patterns such as encrypted communication between a device and a remote host every two minutes
    • Abnormal protocol activity such as gigabytes of DNS traffic or ping requests that do not receive responses or have unusual packet sizes
    • Evidence of traffic on ports associated with C&C traffic using encoded communications and commands
    • Evidence that a user received an email with a suspicious link, visited the site, launched an executable, and established a connection to a system located on the Internet

User and Entity Behavior Analytics (UEBA)

  • UEBA scans indicators from multiple intrusion detection and log sources to identify anomalies
  • often integrated with SIEM platforms
  • identifies malicious behaviors from comparison to baselines
  • tracks user account behavior across different devices and cloud services
  • entity refers to machine accounts
    • e.g., client workstations, virtualized server instances, embedded hardware
  • complex to determine baselines and reduce false positives
    • heavily depend on AI and machine learning
  • tools