Fileless Threat Detection via Code Fragment Inference

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Traditional security measures are inadequate in detecting and mitigating fileless security threats, known as 'living off the land' attacks, which do not rely on traditional portable executables and can evade detection by dynamic behavior analysis tools.

Innovation Solution

A two-stage threat detection and mitigation system comprising a detection engine and an inference engine that identifies suspicious code fragments and reconstructs a rolling map to classify potential attacks, using a combination of prefiltering, feature extraction, and machine learning classifiers to predict and verify connections between fragments.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional antivirus and behavior analysis tools are used, then detection of traditional portable executable threats is improved, but detection of fileless security threats deteriorates

Engineering Contradiction:
Improvedetection capabilityVSAvoidthreat coverage
Core Design Contradiction:
ReliabilityVSAdaptability or versatility

Solution Approach 1:

The patent segments the detection process into two distinct stages: a detection engine that identifies suspicious code fragments using prefiltering and feature extraction, and an inference engine that reconstructs attack scenarios by analyzing connections between fragments. This segmentation allows each component to specialize in specific tasks, improving overall detection capability for both traditional and fileless threats while maintaining reliability.

Inventive Principle:
Principle #1Segmentation

2Measurement precision

If dynamic behavior analysis tools are used, then detection of executable behavior is improved, but detection of fileless attacks deteriorates

Engineering Contradiction:
Improvebehavior detection accuracyVSAvoidfileless threat detection
Core Design Contradiction:
Measurement precisionVSDifficulty of detecting and measuring

Solution Approach 1:

The patent extracts code fragments from various persistent system elements (registry entries, scheduled tasks, WMI entries, etc.) without requiring full executable behavior. The detection engine extracts and analyzes individual suspicious fragments, while the inference engine reconstructs the complete attack picture by connecting these extracted fragments, thereby detecting fileless attacks without relying on traditional dynamic behavior analysis of complete executables.

Inventive Principle:
Principle #2Taking out (Extraction)

3Reliability

If a comprehensive threat detection system is implemented, then detection accuracy is improved, but system complexity deteriorates

Engineering Contradiction:
Improvethreat detection accuracyVSAvoiddetection system complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent implements a dynamic two-stage detection system where the detection engine and inference engine operate in sequence with flexible data flow. The system dynamically adjusts its analysis depth based on suspicion levels, using prefiltering to quickly eliminate benign fragments before applying more complex feature extraction and connection analysis only to suspicious cases, thereby maintaining high accuracy while managing system complexity.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentUS11204998B2Detection and mitigation of fileless security threats
Publication Date: 2021.12.21 MCAFEE LLC
  • US11204998B2 patent drawing
  • US11204998B2 patent drawing
  • US11204998B2 patent drawing

AI summary

There is disclosed in one example a computing apparatus, including: a hardware platform; and a storage medium having stored thereon executable instructions to provide an inference engine configured to: receive a new suspicious fragment object from a protected device; add the new suspicious fragment object to a rolling map configured to provide a temporal snapshot of suspicious fragment objects over a time span; determine a connection between the new suspicious fragment object and an existing suspicious fragment object within the rolling map; apply the connection to a connection map; and operate a map classifier to determine that the connection map represents a probable computer security threat.