ML-Assisted Wireless Positioning for Corner and NLOS Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing wireless positioning techniques face challenges in accurately determining the position of devices in wireless networks due to environmental factors impacting the measurement of positioning reference signals, leading to reduced accuracy.
Innovation Solution
Implementing machine learning (ML) models, such as single-TRP and multi-TRP fingerprinting AI/ML position models, to determine the location of wireless transmit/receive units (WTRUs) by analyzing reference signal received power (RSRP) thresholds and line-of-sight (LOS) or non-line-of-sight (NLOS) conditions, using transmission reference points (TRPs) and positioning reference signals (PRS).
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional positioning techniques are used, then the positioning process is simple, but positioning accuracy is reduced due to environmental factors impacting reference signal measurements
Solution Approach 1:
The patent introduces machine learning models as an intermediary between the raw reference signal measurements and the final position determination. These ML models process the RSRP measurements and environmental indicators to compensate for environmental distortions, thereby improving positioning accuracy without requiring changes to the fundamental positioning infrastructure
Solution Approach 2:
The patent transforms the positioning approach by changing from direct geometric calculation based on signal measurements to a data-driven approach using machine learning models. The system learns to map signal characteristics and environmental indicators to accurate position estimates, effectively changing the computational parameters and methods used in positioning
2Measurement precision
If machine learning models are implemented to improve positioning accuracy, then positioning precision is enhanced, but the complexity of the positioning system increases
Solution Approach 1:
The patent segments the positioning system into distinct functional components: signal measurement modules, environmental indicator detection modules, machine learning model modules, and position determination modules. This segmentation allows each component to be optimized independently and facilitates easier integration and maintenance of the complex ML-based positioning system
3Measurement precision
If environmental factors are considered in positioning, then positioning accuracy is improved, but the time required for position determination increases
Solution Approach 1:
The patent performs preliminary actions by pre-training machine learning models with extensive environmental data and pre-processing environmental indicators during signal measurement. This allows the system to quickly apply pre-learned patterns and pre-processed data during actual positioning operations, reducing the time required for real-time position determination while maintaining high accuracy
Data Source
AI summary
Methods, devices, and systems for machine learning (ML)-assisted position determination are disclosed. Information is received which indicates artificial intelligence/machine learning (AI/ML) models for determining position. Information is received which indicates transmission reference points (TRPs) (502, 504) associated with corners. Information is received which indicates a reference signal received power (RSRP). The TRPs associated with corners include a first TRP. It is determined that the WTRU is located in a corner based on an RSRP of a positioning reference signal (PRS) (506) received from the first TRP being above an RSRP threshold. Position information is determined based on an AI/ML position model and the determination that the WTRU is located in the corner. Information indicating the position of the WTRU is transmitted.