AI/ML Positioning Data Processing with Virtual Reference UEs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In scenarios with a limited number of Positioning Reference Units (PRUs), the data requirements for training, monitoring, and updating AI/ML models for accurate positioning cannot be met, posing a challenge in Indoor Factory Dense Hall (InF-DH) environments.
Innovation Solution
A data processing method and apparatus that utilizes network devices to send indication information to determine user equipment, receive and process signals from user equipment, and perform training, monitoring, or updating of AI/ML models to determine location-related information, leveraging channel observation results and other location-related information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If AI/ML-based positioning is implemented, then positioning accuracy is improved, but data requirements cannot be met when only a limited number of PRUs are available
Solution Approach 1:
The patent enables ordinary user equipment to serve multiple functions: acting as positioning targets for location determination, serving as reference devices for providing positioning data, and participating in model training. This multi-functionality allows the system to gather sufficient training data without requiring additional dedicated PRUs, thus resolving the contradiction between positioning accuracy and data availability.
Solution Approach 2:
The patent creates virtual copies of reference devices by designating certain user equipment as reference devices. These virtual reference devices provide positioning data similar to actual PRUs, enabling the system to generate sufficient training data through software-defined roles rather than physical hardware multiplication.
2Device complexity
If a limited number of PRUs are used, then device complexity is reduced, but data requirements for training and updating AI/ML models cannot be met
Solution Approach 1:
The system enables user equipment to self-organize into reference devices based on pre-configured criteria. Devices automatically determine whether they meet the conditions to serve as reference devices and provide positioning data accordingly, eliminating the need for complex centralized management of PRUs while ensuring sufficient data collection for AI/ML model training.
Solution Approach 2:
The patent shifts the system from a hardware-centric approach (relying on physical PRUs) to a software-defined approach where any user equipment can become a reference device. This dimensional shift from physical hardware to virtual functionality expands data sources without increasing physical device complexity.
3Quantity of substance
If more PRUs are deployed to meet data requirements, then data sufficiency is improved, but system complexity and cost increase
Solution Approach 1:
The patent implements a dynamic reference device selection mechanism where user equipment can transition between being a positioning target and a reference device based on real-time conditions and pre-configured criteria. This dynamic approach allows the system to adaptively utilize available devices for data collection without requiring a fixed, over-provisioned infrastructure.
Solution Approach 2:
The network device acts as an intermediary that coordinates between positioning targets and reference devices, managing the selection and data collection process centrally without requiring direct peer-to-peer communication between all devices. This intermediary approach simplifies system management while enabling complex multi-device interactions.
Data Source
Figure 1~2
Figure 3~5
Figure 6~7
AI summary
The present disclosure provides a data processing method and apparatus and relates to the field of communication technologies. The method includes: sending, by a network device, first indication information, wherein the first indication information is used to determine first user equipment (UE); receiving, by the network device, first information and/or a first signal sent by a second user equipment, wherein the first information and/or the first signal is sent by the second user equipment when the second user equipment determines that the second user equipment serves as the first user equipment; sending, by the network device, the received first information and/or first signal to a first device, or, according to the received first information and/or first signal, performing, by the network device, one or more of the following: training of a first model, performance monitoring of the first model, and updating of the first model. The first model is used to determine UE location-related information.