AI/ML Data Collection for Wireless Positioning Model Adaptation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional wireless positioning methods, especially in non-line-of-sight environments, suffer from poor accuracy due to large measurement errors, and existing AI/ML models lack effective real-time data collection and management protocols, leading to insufficient performance in complex wireless communication environments.
Innovation Solution
A data collection method and apparatus are introduced to manage AI/ML model data collection and configuration between network entities and terminals, optimizing wireless positioning by real-time data exchange and lifecycle management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional wireless positioning methods based on channel measurement are used, then positioning can be performed in various wireless environments, but positioning accuracy deteriorates significantly in non-line-of-sight environments due to large measurement errors
Solution Approach 1:
The patent introduces AI/ML models as an intermediary between traditional channel measurement methods and positioning results. These models process channel state information and measurement data to compensate for errors in non-line-of-sight environments, thereby maintaining positioning accuracy while preserving the broad applicability of traditional methods
Solution Approach 2:
The patent transforms the positioning approach by changing from direct channel measurement parameters to AI/ML model parameters that capture environmental characteristics. The system adapts model parameters based on wireless environment conditions, enabling accurate positioning in both line-of-sight and non-line-of-sight scenarios
2Measurement precision
If AI/ML models are deployed for wireless positioning, then positioning accuracy is improved, but model generalization performance deteriorates in complex and variable wireless communication environments
Solution Approach 1:
The patent implements dynamic model selection and adaptation mechanisms where the AI/ML system can switch between different models or adjust model parameters based on current wireless environment conditions. This dynamic approach allows the system to maintain high accuracy across varying environments without requiring a single universal model
Solution Approach 2:
The patent segments the wireless environment into different scenarios or zones with distinct characteristics, and applies specialized AI/ML models for each segment. This segmentation strategy improves generalization by training models on specific environmental conditions rather than attempting to create a single model for all possible scenarios
3Reliability
If real-time data collection and model management are implemented, then AI/ML model performance is optimized, but system complexity increases due to additional signaling procedures and data management requirements
Solution Approach 1:
The patent designs the data collection and management system to serve multiple functions simultaneously: collecting training data, monitoring model performance, updating models, and managing multiple AI/ML models. This multi-functionality reduces overall system complexity by consolidating what would otherwise require separate systems for each function
Data Source
AI summary
A data generation apparatus includes: a transmitter configured to transmit request information for collecting data to a model deployment apparatus; and a receiver configured to receive AI/ML model-related information from the model deployment apparatus; wherein the transmitter is further configured to transmit data to the model deployment apparatus according to the AI/ML model-related information.