IoT Device Classification via Context-Normalized Telemetry Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
As the Internet of Things (IoT) grows, existing device classification methods struggle to accurately identify device types in computer networks due to the increasing number and variety of IoT devices, often misclassifying devices due to environmental and local context noise in network traffic.
Innovation Solution
A device classification service that assigns endpoint devices to context groups, forms context summary feature vectors, normalizes telemetry feature vectors by removing environmental context, and uses these normalized vectors for device type classification, employing machine learning-based classifiers to improve accuracy and generalize across different environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional device classification methods are used in IoT networks, then device identification can be performed, but classification accuracy deteriorates due to environmental and local context noise in network traffic
Solution Approach 1:
The patent extracts and removes environmental context features from network traffic data before classification. The system identifies and separates device-specific features from environmental noise, retaining only the relevant device characteristics for classification. This extraction process eliminates the harmful effect of environmental context noise on classification accuracy.
Solution Approach 2:
The patent transforms the input data parameters by applying normalization and feature selection techniques. It changes the parameter representation from raw network traffic data containing environmental noise to normalized feature vectors that emphasize device-specific characteristics. This parameter transformation improves measurement precision by altering the data representation to be more suitable for accurate classification.
2Reliability
If device classification is performed using raw network traffic data, then classification can be conducted, but misclassification rates increase due to environmental noise
Solution Approach 1:
The system extracts device-specific features from the mixed network traffic data by removing environmental context. This extraction process increases classification reliability by isolating the relevant device characteristics while eliminating confounding environmental factors that cause misclassification.
Solution Approach 2:
The patent applies different processing treatments to different features in the network traffic data. Device-specific features are preserved and emphasized, while environmental features are removed or downweighted. This local quality approach ensures that the classification process focuses on the most reliable indicators of device type.
3Measurement precision
If context-specific classification models are created for different environments, then local accuracy may improve, but system complexity increases
Solution Approach 1:
The patent creates a universal classification system that works across different environments by removing environment-specific features. A single classification model can be deployed universally without needing separate models for each environment, reducing system complexity while maintaining accuracy through context-agnostic feature extraction.
Solution Approach 2:
Instead of adapting the classification model to each environment (which increases complexity), the patent inverts the approach by adapting the data to be environment-independent. The normalization process transforms local environmental variations into a unified representation that a single model can handle effectively.
Data Source
AI summary
In one embodiment, a device classification service assigns a set of endpoint devices to a context group. The device classification service forms a context summary feature vector for the context group that summarizes telemetry feature vectors for the endpoint devices assigned to the context group. Each telemetry feature vector is indicative of a plurality of traffic features observed for the endpoint devices. The device classification service normalizes a telemetry feature vector for a particular endpoint device using the context summary feature vector. The device classification service classifies, using the normalized telemetry feature vector for the particular endpoint device as input to a device type classifier, the particular endpoint device as being of a particular device type.