IoT Edge Classification via Remote Feature Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Computing devices in the Internet of Things (IoT) environments face challenges in processing complex data due to limited processing power, making it difficult to efficiently handle large computing tasks.
Innovation Solution
The solution involves deploying an encoder with high computing resources in a remote IoT cloud device to extract features from sample data, while a classifier with limited resources is deployed on edge or local devices to classify the data, optimizing the use of computing power and reducing network transmission requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If computing devices in IoT process complex sample data locally, then classification accuracy can be maintained, but processing efficiency deteriorates due to limited computing power
Solution Approach 1:
The patent divides the neural network model into two separate components: an encoder for feature extraction and a classifier for category determination. The encoder is deployed on remote devices with abundant computing resources, while the classifier is deployed on local edge devices with limited resources. This segmentation allows each component to be optimized for its specific function and deployment environment, resolving the contradiction between maintaining accuracy and improving processing efficiency.
2Reliability
If the entire neural network model is deployed on local edge devices, then data privacy is protected, but processing capability deteriorates due to resource constraints
Solution Approach 1:
The patent introduces an intermediate representation (feature vectors) that bridges the remote encoder and local classifier. Instead of transmitting raw sample data or complete model predictions, the encoder extracts feature vectors that capture essential information while being compact and privacy-preserving. These feature vectors serve as an intermediary that enables distributed processing without requiring full model deployment on resource-constrained devices.
3Measurement precision
If complex feature extraction is performed locally, then model performance can be optimized, but energy consumption increases beyond sustainable levels
Solution Approach 1:
The patent extracts the computationally intensive feature extraction function from the local edge device and places it on remote devices with abundant resources. The encoder performs complex feature extraction operations remotely, then transmits only the extracted feature vectors to the local classifier. This extraction of heavy computation from the energy-constrained local device dramatically reduces energy consumption while maintaining model performance.
Data Source
AI summary
The present disclosure relates to a method, a device, and a program product for processing sample data in an Internet of Things environment. A method in one embodiment includes: receiving features of the sample data from an encoder deployed in a remote device in the Internet of Things environment; acquiring a category probability corresponding to the sample data based on a classifier deployed in a local device in the Internet of Things environment and the features; and classifying the sample data to a predetermined category in response to determining that the category probability satisfies a first threshold condition. Further, a corresponding device and a corresponding program product are provided. With example implementations of the present disclosure, computing resources of devices in an Internet of Things environment can be fully utilized to process sample data.