Network Service Detection with Segmented Multi-Layer Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing network traffic analysis methods struggle to accurately identify multiple types of services in a network traffic stream, especially with the increasing use of encrypted traffic, necessitating a more reliable method to segregate and categorize traffic patterns using machine learning.
Innovation Solution
A network connected device that decomposes network traffic into data flows based on source and destination information, filters these flows using machine learning, and employs a multi-layer model to determine and sub-categorize service types, incorporating sensor information for enhanced accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If network traffic is analyzed using traditional methods, then the analysis process is simple, but the accuracy of identifying multiple service types is insufficient
Solution Approach 1:
The patent segments network traffic into multiple data flows based on source and destination information, allowing each flow to be analyzed independently for service type identification. This segmentation enables the system to handle multiple service types simultaneously while maintaining analysis accuracy without overwhelming complexity
Solution Approach 2:
The patent introduces a multi-layer machine learning model that adds dimensional depth to the analysis process. The multi-layer architecture processes traffic characteristics through multiple levels of abstraction, transforming the analysis from a single-dimension approach to a multi-dimensional one, thereby improving identification accuracy while managing complexity through structured processing
2Reliability
If all data flows are processed for service type identification, then comprehensive coverage is achieved, but the processing time and computational resources increase
Solution Approach 1:
The patent extracts and processes only the most relevant traffic characteristics and data flows that are critical for service type identification. By selecting key features rather than processing all raw traffic data, the system achieves comprehensive identification coverage while significantly reducing processing time and computational resource requirements
Solution Approach 2:
The patent applies partial processing by focusing on representative samples and key traffic patterns rather than exhaustively analyzing every single packet. This approach provides sufficient accuracy for reliable service type identification while avoiding the excessive time and computational costs of complete analysis
Data Source
AI summary
A method and apparatus for detecting network service types by grouping together applications that have similar latency requirement and data characteristics to form a service type. Machine learning algorithms may be used to detect the traffic pattern in the traffic stream by using features extracted from packet information and optionally additional sensor information. Subsequently, the output of the machine learning module may go through a post-processing process that may employ different techniques to use current prediction and past predictions to make a final decision about the network service type.