Privatized Communication Data for Secure AI Dataset Generation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In millimeter wave wireless communication, sensitive information of a second device is prone to leakage on a first device during data collection, leading to poor information security.
Innovation Solution
A dataset generation method where a first device receives first information from a second device, including a first identifier and data information processed in a target manner, generating a dataset for AI model processing that includes model inference, training, or monitoring, while ensuring the first device does not recover the original sensitive information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the first device receives information related to device communication from the second device to implement data collection, then the data collection capability is improved, but the information security deteriorates due to sensitive information leakage
Solution Approach 1:
The patent introduces an intermediary processing mechanism where the second device processes the communication information locally before transmission. The first device receives only the processed results (first information) rather than the original sensitive data, acting as an intermediary that prevents direct exposure of sensitive information while still enabling data collection for AI model training.
Solution Approach 2:
The patent segments the data collection process into two distinct stages: (1) the second device processes and anonymizes the sensitive communication information locally, and (2) the first device receives and uses only the anonymized first information. This segmentation ensures that sensitive information remains confined to the second device while still enabling the first device to perform its data collection function.
2Measurement precision
If the first device processes original communication information directly, then the AI model training accuracy is improved, but the information security deteriorates due to exposure of sensitive data
Solution Approach 1:
The patent applies preliminary action by having the second device process and anonymize the communication information before it is transmitted to the first device. This preprocessing step removes sensitive elements in advance, allowing the first device to work with safe, anonymized data that retains sufficient information quality for AI model training without exposing sensitive information.
Solution Approach 2:
The patent creates a copy of the communication information that has been processed and anonymized by the second device. The first device receives and uses this copied, sanitized version of the data rather than the original sensitive information, maintaining training utility while eliminating security risks associated with direct access to sensitive data.
Data Source
Figure 1a~1b
Figure 2~3
Figure 4~6
AI summary
This application discloses a dataset generation method, an information sending method, apparatuses, and related devices, and pertains to the field of communication technology. The dataset generation method of an embodiment of this application includes: receiving, by a first device, first information from a second device, where the first information includes a first identifier and data information, the data information being information obtained by processing information related to device communication in a target processing manner and the first identifier being used to indicate the target processing manner; and generating, by the first device, a dataset based on the first identifier, where the dataset includes data information associated with the first identifier, the dataset is used for artificial intelligence AI model processing, and the AI model processing includes at least one of model inference of AI models, model training of AI models, and model monitoring of AI models.