Beam Management Data Collection Through Privacy-Preserving Features
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In AI-based beam management, the exposure of sensitive beam or antenna information leads to low accuracy in model-based predictions due to the inability to obtain detailed transmit and receive beam information simultaneously, compromising privacy and model accuracy.
Innovation Solution
A data collection method where a first device constructs a sample dataset with sensitive information, uses a model for feature extraction, and sends processed information to a second device, allowing the second device to determine target sample data without exposing its own sensitive information, thereby improving data integrity and model training accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If detailed information of transmit beam and receive beam are obtained simultaneously, then accuracy of model-based beam prediction is improved, but sensitive beam or antenna information is exposed
Solution Approach 1:
The patent introduces a first model as an intermediary component that processes sensitive beam information through feature extraction. This mediator transforms raw sensitive information into processed features that can be shared for model training without exposing the original sensitive data, thereby enabling accurate predictions while protecting privacy.
Solution Approach 2:
The patent segments the beam information processing into distinct components: sensitive information collection, feature extraction through the first model, and shared feature transmission. This segmentation allows the system to separate sensitive data from processed features, enabling collaboration between devices without direct exposure of sensitive information.
2Reliability
If sensitive information is shared between devices, then model training accuracy is improved, but device privacy is compromised
Solution Approach 1:
The first model acts as a privacy-protecting intermediary that extracts features from sensitive information before sharing. This mediator ensures that only processed features are transmitted to the second device, maintaining model training accuracy while preventing direct access to sensitive device information.
Solution Approach 2:
The patent creates a processed copy of sensitive information through feature extraction. Instead of sharing original sensitive data, the system transmits extracted features that replicate the necessary information patterns for model training while being irreversibly detached from the original sensitive source.
Data Source
AI summary
A data collection method, a terminal, and a network-side device are provided. The data collection method includes: constructing a first sample dataset, where first sample data in the first sample dataset includes sensitive information of the first device; determining a first output of a first model based on the first sample dataset, where the first model is used for performing feature extraction on the sensitive information of the first device; and sending first information to a second device. The first information is determined based on the first output of the first model, and is used for the second device to determine target sample data that includes first target information and second target information. The first target information is information determined based on a first output corresponding to the first sample data. The second target information is determined based on second sample data including sensitive information of the second device.