Malicious Communication Pattern Extraction for Malware Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for detecting malware-infected terminals often result in erroneous detection due to malware communicating through normal or non-standard channels, leading to increased operational costs and time spent on manual analysis.
Innovation Solution
A malicious communication pattern extraction device that calculates statistical values for communication patterns from both malware and normal traffic logs, compares their frequencies, and sets thresholds to minimize erroneous detection while ensuring effective malware detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If simple URL extraction is used to detect malware communication, then the detection process is simple, but erroneous detection increases and detection accuracy decreases
Solution Approach 1:
The patent changes the detection parameters from simple URL extraction to multi-dimensional communication pattern analysis including protocol types, port numbers, payload characteristics, and temporal patterns. This parameter transformation enables accurate differentiation between malware communication and normal traffic while maintaining operational simplicity through automated pattern matching.
Solution Approach 2:
The patent segments the communication detection into multiple independent analysis dimensions (protocol analysis, port number analysis, payload analysis, temporal analysis). Each dimension independently evaluates specific communication characteristics, and the results are combined to make final detection decisions, reducing erroneous detection while maintaining simplicity.
2Adaptability or versatility
If malware communication patterns are expanded to include normal communication, then more communication types are detected, but detection specificity decreases
Solution Approach 1:
The patent applies local quality analysis by examining specific local characteristics of each communication pattern (unique port numbers, specific protocol combinations, distinctive payload signatures, particular temporal intervals). These local characteristics serve as fingerprints that differentiate malware communication from normal communication even when both use standard protocols, enabling high specificity while maintaining broad coverage.
Solution Approach 2:
The patent creates composite detection patterns by combining multiple communication characteristics (protocol + port number + payload + timing) into a unified detection model. This composite approach allows the system to recognize malware communication across diverse patterns while maintaining high specificity through the combination of multiple discriminative features.
3Measurement precision
If manual analysis is performed for erroneous detection, then detection accuracy can be improved, but operational time and cost increase
Solution Approach 1:
The patent implements self-service detection through automated communication pattern extraction and classification. The system automatically learns normal communication patterns from baseline data, automatically identifies deviations as potential malware behavior, and automatically generates detection rules. This eliminates the need for manual analysis of erroneous detections while maintaining high accuracy through continuous automated learning and adaptation.
Solution Approach 2:
The patent incorporates feedback mechanisms where detection results are continuously fed back into the system to refine and update communication patterns. The system learns from both confirmed malware detections and false positives, automatically adjusting its pattern recognition algorithms to improve accuracy over time without requiring manual intervention or time loss.
Data Source
AI summary
A malicious communication pattern extraction device includes: a statistical value calculation unit that calculates a statistical value for an appearance frequency of each of plural communication patterns, from a traffic log obtained from traffic caused by malware, and a traffic log obtained from traffic in a predetermined communication environment; a malicious list candidate extraction unit that compares between the appearance frequency of the traffic logs for each of the communication patterns, based on the calculated statistical value, and extracts the communication pattern as the malicious communication pattern when a difference between both of the appearance frequencies is equal to or more than a predetermined threshold; and a threshold setting unit that sets a threshold so that an erroneous detection rate probability of erroneously detecting the traffic caused by malware and a detection rate probability of detecting the traffic caused by malware is equal to or more than a certain value.