Traffic Analysis Apparatus Using Information Amount Calculation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Telecommunications operators face challenges in accurately inferring the state of communication traffic due to limitations in available information and the complexity of time series data, leading to erroneous recognition and decreased inference accuracy, especially with increasing encrypted traffic and diverse communication needs.
Innovation Solution
A traffic analysis apparatus and method that calculates information amounts of time series data and selects relevant data items for analysis, using techniques like sample entropy to identify suitable patterns, and creates probabilistic state transition models using hidden Markov models to improve inference accuracy while reducing computational complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If deep packet inspection (DPI) is used to inspect inside of traffic, then communication quality control is improved, but the technology becomes limited due to encrypted traffic and privacy laws
Solution Approach 1:
The patent extracts only the necessary features from traffic data for analysis rather than inspecting the complete packet content. By using feature extraction techniques, the system obtains sufficient information for traffic analysis without requiring deep packet inspection, thereby maintaining effectiveness with encrypted traffic while reducing computational overhead.
Solution Approach 2:
The patent introduces an intermediary layer of traffic analysis that operates on metadata and observable traffic characteristics rather than directly inspecting packet contents. This intermediary approach allows the system to analyze encrypted traffic by examining patterns in timing, size, and flow characteristics without decrypting the actual content.
2Measurement precision
If the number of control target parameters is increased to achieve fine communication control, then communication quality is improved, but the complexity of analysis increases
Solution Approach 1:
The patent segments the traffic analysis process into distinct modules: traffic data acquisition, feature extraction, pattern recognition, and state inference. By dividing the complex analysis task into smaller, specialized segments, the system can handle multiple control parameters efficiently without overwhelming complexity in any single component.
Solution Approach 2:
The patent transforms raw traffic parameters into derived features and patterns that capture essential communication quality characteristics. By changing the parameter representation from raw packet data to meaningful features (such as traffic flow patterns, timing characteristics, and size distributions), the system achieves fine-grained control with reduced analytical complexity.
3Measurement precision
If multiple items of time series data are used to infer application state, then inference accuracy is improved, but the calculation amount increases
Solution Approach 1:
The patent extracts only the most relevant features from multiple time series data sources rather than processing all available data. By identifying and extracting key features that have the highest correlation with application state, the system maintains high inference accuracy while significantly reducing the computational burden of analyzing multiple data streams.
Solution Approach 2:
The patent applies partial action by selectively analyzing only the subset of time series data that provides the most valuable information for inference. Rather than processing all available data equally, the system focuses computational resources on the most informative features, achieving effective inference with reduced calculation requirements.
Data Source
AI summary
A traffic analysis apparatus includes an information amount calculation part that calculates information amounts of a plurality of items of time series data relating to communication traffic and an input information selection part that selects at least one item of time series data based on the information amounts of the plurality of items of time series data.