UL-RS Monitoring Configuration for AI Positioning Model Validation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing wireless communication systems face challenges in accurately monitoring artificial intelligence machine learning (AIML) models for network-assisted position estimation due to the lack of scalable ground truth methods, particularly when frequent monitoring is required to adapt to changes in environment and user equipment mobility.
Innovation Solution
Implementing a distinct uplink reference signal (UL-RS) configuration for AIML model monitoring, separate from the configuration used for position estimation sessions, to validate, invalidate, or update the AIML model, even when no active position estimation session is ongoing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single UL-RS configuration is used for both position estimation sessions and AIML model monitoring, then device complexity is reduced, but positioning accuracy and model adaptability deteriorate because continuous validation cannot be performed outside active sessions
Solution Approach 1:
The patent segments the UL-RS configuration into two distinct configurations: one for position estimation sessions and another for AIML model monitoring. This segmentation allows each configuration to be optimized for its specific purpose, enabling continuous model validation without being constrained by session-based timing, thereby improving positioning accuracy while maintaining manageable complexity through structured separation of functions.
Solution Approach 2:
The patent creates a universal UL-RS configuration for AIML model monitoring that can operate independently of position estimation session states. This multi-functional approach allows the monitoring configuration to serve continuous model validation purposes while the session configuration handles active positioning, resolving the contradiction by enabling both functions to coexist with optimized performance.
2Use of energy by moving object
If UL-RS is transmitted only during active position estimation sessions, then energy consumption is reduced, but AIML model adaptability to environmental changes and mobility deteriorates due to lack of continuous monitoring
Solution Approach 1:
The patent segments the monitoring function from session-based operations by creating a dedicated UL-RS configuration for AIML model monitoring that operates continuously or independently of session states. This segmentation enables energy-efficient selective activation where monitoring can occur at optimized intervals rather than continuously during all sessions, balancing energy consumption with continuous model adaptability to environmental changes and mobility.
Solution Approach 2:
The patent implements periodic or scheduled UL-RS transmission for AIML model monitoring at configured intervals regardless of session state. This periodic action ensures continuous model validation and adaptability while consuming energy only at necessary intervals, resolving the contradiction between energy efficiency and continuous monitoring requirements.
3Measurement precision
If frequent UL-RS transmission is implemented for continuous AIML model monitoring, then positioning accuracy and model adaptability improve, but network load and signal interference increase
Solution Approach 1:
The patent segments the UL-RS transmission into two distinct configurations with different transmission parameters: one for position estimation sessions and another for AIML model monitoring. The monitoring configuration can be optimized with lower transmission frequency, reduced power, or selective activation based on model drift detection, thereby maintaining positioning accuracy through continuous validation while reducing overall network load compared to frequent transmissions during every session.
Solution Approach 2:
The patent changes key transmission parameters for the monitoring configuration compared to session-based transmission, such as reducing transmission frequency, adjusting power levels, or implementing event-triggered transmission based on environmental change detection. These parameter changes enable continuous model monitoring for accuracy while managing network load by transmitting only when necessary for validation.
Data Source
AI summary
Aspects of the disclosure are thereby directed to an uplink reference signal (UL-RS) configuration associated with monitoring of an artificial intelligence machine learning (AIML) model for network-assisted position estimation (e.g., Case 2a, 2b, 3a and/or 3b for AIML model positioning use cases for 3GPP). In an aspect, the UL-RS configuration associated with monitoring of the AIML model may be different than a UL-RS configuration utilized for position estimation of the UE. In further aspects, the UL-RS configuration associated with monitoring of the AIML model may be utilized even if there is no active position estimation session of the UE. Such aspects may provide various technical advantages, such as improving positioning performance associated with AIML models (e.g., by utilizing the AIML model monitoring to validate/invalidate the AIML model, switch between AIML models, update the AIML model, etc.) while not being limited to UL-RS configurations for position estimation sessions.